The Artificial Intelligence Opportunity: A Camel to Cars Moment

Two_Focus_AreasOver the last couple years, I’ve spent an increasing amount of time diving into the possibilities Deep Learning (DL) offers in terms of what we can do with Artificial Intelligence (AI). Some of these possibilities have already been realized (more on this later in the post). And, I could not be more excited to see them out in the world.

Through it all, I’ve felt there are a handful of breath-taking realities that most people are not grasping when it comes to an AI-Powered world. Why the implications are far deeper for humanity than we imagine. Why in my areas of expertise, marketing, sales, customer service and analytics, the impact will be deep and wide. Why is this not yet another programmatic moment. Why the scale at which we can (/have to) solve the problems is already well beyond the grasp of the fundamental strategy most companies follow: We have a bigger revenue opportunity, but we don’t know how to take advantage? Let’s buy more hamster wheels, hire more hamsters and train them to spin faster!

Today I want shed some light on these whys, and a bit more. My goal is to try to cause a shift in your thinking, to get you to take a leadership role in taking advantage of this opportunity both at a personal and professional level.

I’ve covered AI earlier: Artificial Intelligence: Implications On Marketing, Analytics, And You. You’ll learn all about the Global Maxima, definitions of AI/ML/DL, and the implications related to the work we do day to day. If you’ve not read that post, I do encourage you to do so as it will have valuable context.

In this post, I’ve organized my thoughts into these six clusters:

There is a deliberate flow to this post, above. If you are going to jump around, it is ok, but please be sure to read the section below first. You won’t regret it.

Ready to have your mind stretched? Let’s go!

What’s the BFD?

I’m really excited about what’s in front of us. When I share that excitement in my keynotes or an intimate discussion with a company’s board of directors, I make sure I stress two especially powerful concepts that I have come to appreciate about the emerging AI solutions: Collective Continuous Learning + Complete Day One Knowledge.

They are crucial in being able to internalize the depth and breadth of the revolution, and why we strengths AI brings are a radical shift beyond what humans are capable of.

The first eye-opening learning for me came from the Google Research team’s post on Learning from Large-Scale Interaction.

Most robots are very robotic because they follow a sense-plan-act paradigm. This limits the types of things they are able to do, and as you might have seen their movements are deliberate. The team at Google adopted the strategy of having a robot learn own its own (rather than programming it with pre-configured models).

The one-handed robots in this case had to learn to pick up objects.

Initially the grasping mechanism was completely random – try to imagine a baby who barely knows they even have a hand at the end of their shoulder. Hence, you’ll see in the video below, they rarely succeed at the task at hand. ;)

At the end of each day, the data was collected and used to train a deep convolutional neural network (CNN), to learn to predict the outcome of each grasping motion. These learnings go back to the robot and improve its chances of success.

Here’s the video…


(Play on YouTube)

It took just 3,000 robot-hours of practice to see the beginnings of intelligent behavior.

What’s intelligent behavior of a CNN powered one-handed robot?

Among other things, being able to isolate one object (a stapler) to successfully pick-up a Lego piece. You’ll see that at 15 seconds in this video…


(Play on YouTube)

Or, learning how to pick up different types of objects (a dish washing soft sponge, a blackboard eraser, or a water glass  etc.).

I felt a genuine tingling sensation just imagining a thing not knowing something and it being able to simply learn. I mean pause. Just think about it. It started from scratch – like a baby – and then just figured it out. Pretty damn fast. It truly is mind-blowing.

There were two lessons here. The first related to pure deep learning and its amazingness, I was familiar with this one. The second was something new (for me). This experiment involved 14 one-handed robot arms. While not a massive number, the 14 were collectively contributing data from the start – with their many failures. The end of day learnings by the convolutional neural network were using all 14. And, the next day, all 14 started again with this new level of collective wisdom.

For a clear way for me to capture this lesson, I call this Collective Learning.

It is very powerful.

Think of 14 humans learning a new task. Peeling an apple. Or, laying down track for a railroad. Or, programming a new and even more frustrating in-flight entertainment menu for Air Canada (who have the worst one known to mankind).

Every human will do it individually as well as they can – there will be the normal bell curve of competency. It is entirely possible, if there are incentives to do so, that the humans who are better in the group will try to teach others. There will be great improvement if the task is repetitive and does not require imagination/creativity/intrinsic intelligence. There might be a smaller improvement if the task is not repetitive and requires imagination/creativity/intrinsic intelligence.

In neither case will there be anything close to Collective Learning when it comes to humans.

Humans also do not posses this continuous closed loop: Do something. Check outcome (success or failure). Actively learn from either, improve self. Do something better the next time.

Collective Continuous Learning. An incredible advantage that I had simply not thought through deeply enough.

Here’s the second BFD.

Machine Learning is already changing lots of fields, the one I’m most excited about is what’s happening in healthcare. From the ability to speed up discovery of new medicines to the unbelievable speed with which Machine Learning techniques are becoming particularly adept at diagnosis (think blood reports, X-rays, cancers etc.). 

An example I love. 415 million diabetic patients worldwide are at risk of Diabetic Retinopathy (DR) – the fastest growing cause of blindness. If caught early, the disease is completely treatable. The problem? Medical specialists capable of detecting DR are rare in many parts of the world where diabetes is prevalent.

Using a dataset of 128,000 images Google’s  Accelerated Science Team trained a deep neural network to detect DR from retinal photographs. The results delivered by the algorithm (black curve) were slightly better than expert ophthalmologists (colored dots)…

diabetic_retinopathy_deep_learning_algorithm_results

Specifically the algorithm has a F-score of 0.95 and the median F-score of the eight expert ophthalmologists was 0.91.

As richer datasets become available for the neural network to learn from, as 3D imaging technology like Optical Coherence Tomography becomes available all over the world to provide more detailed view of the retina, just imagine how transformative the impact will be.

Literally millions upon millions of people at risk of blindness will have access to AI-Powered technology that can create a different outcome for their life  – and their families.

#omg

A recent incredible article on this topic is in my beloved New Yorker magazine: A.I. VERSUS M.D. You *should* read it. I’ll jump to a part of the article that altered my imagination of possibilities.

An algorithm created by Sebastian Thrun, Andre Esteva and Brett Kuprel can detect keratinocyte carcinoma (a type of skin cancer) by looking at images of the skin (acne, a rash, mole etc.). In June 2015 it got the right answer 72% of the time, two board-certified dermatologists got the right answer for the same images 66% of the time.

Since then, as they outlined in their report published in the prestigious journal Nature, the algorithm has gotten smarter across even more skin cancer types – and consistently performs better than dermatologists.

Most cancers are fatal because they are detected too late, just imagine the transformative impact of this algorithm sitting in the cloud easily accessible to all humanity via their five billion smartphones. This dream come true: low-cost universal access to vital diagnostic care.

Oh, and here’s a profoundly under-appreciated facet of all this. These health algorithms (including and beyond the one above), are incredible at corner cases, the rare long-tail anomalies. They don’t forget what they have seen once or “rarely.”

This is just a little bit of context for the key point.

A dermatologist in a full-time practice will see around 200,000 cases during her/his lifetime. With every case she sees, she’ll ideally add to her knowledge and grow her diagnostic skills.

Our very human problem is that every new dermatology resident starts almost from scratch. Some textbooks might be updated (while comfortably remaining a decade of more behind). Some new techniques – machines, analytical strategies – might be accessible to the resident. But, the depth and breadth of knowledge acquired by the dermatologist at the end of her career with 200k cases, is almost completely inaccessible to the new resident. Even if they do a residency at an hospital or with a old dermatologist, a newly minted dermatologist will only be a little better than when the old one left school.

Consider this instead: The algorithm above processed 130,000 cases in three months! And every day it will get smarter as it’ll have access to the latest (and more) data. Here though is the magical bit. Every single new algorithm we bring online will have total access to all knowledge from previous algorithms! It’s starting point will be, what I call, Complete Day One Knowledge.

As it gets more data to learn from, as it has access to more compute power, it will get smarter and build upon that complete knowledge. The next version of the algorithm will start with this new high mark.

There is nothing equivalent to Complete Day One Knowledge when it comes to humans.

Combine having Complete Day One Knowledge with Collective Continuous Learning (networked hardware or software all learning at the same time) and it should take you five seconds to realize that we are in a new time and place.

Whatever form AI takes, it will always have access to complete knowledge and through the network each instance will make all others smarter every single instance/moment of its existence.

Humans simply can’t compete.

That’s the BFD.

Stop. Think. If you disagree even slightly, scroll back up and read the post again.

It is imperative that you get this not because of what will happen in 10 years, but what is happening today to the job you have. If you still disagree, scroll down and post a comment, I would love to hear your perspective and engage in a conversation.

deepmind_relational_reasoning

Bonus 1: There is an additional valuable lesson related to open-loop grasp selection and blindly executing it vs. incorporating continuous feedback (50% reduction in failure rates!). The two videos are worth watching to see this in action.

Bonus 2: While we are on the subject of objects… Relational reason is central to human intelligence. Deepmind has had recent success in building a simple neural network module for relational reasoning. This progress is so very cool. Additionally, I was so very excited about the Visual Interaction Network they built to mimic a human’s ability to predict. (If you kick a ball against the wall, your brain predict what will happen when the ball hits the wall.) The article is well worth reading: A neural approach to relational reasoning. Success here holds fantastic possibilities.

Wait. So are we “doomed”?

It depends on what you mean by doomed but: Yes. No. Yes, totally.

Artificial Intelligence will hold a massive advantage over humans in the coming years.

In field after field due to Collective Continuous Learning and Complete Day One Knowledge (not to mention advances in deep learning techniques and hardware :)), AI will be better at frequent high-volume tasks.

Hence, the first yes.

Neuralink at the moment is a concept (implantable brain-computer interface). But many experts (like Ray Kurzweil) believe some type of connection between our human brain and “intelligence, data, compute power in the cloud” will be accessible to humans.

I humbly believe that when that happens, over the next few decades (think 2050), humans could get to parity with AI available at that time. We might even have an advantage for some time (if only because I can’t let go of the thought that our brains are special!).

Hence, the no.

As we head towards the second half of the current century, AI will regain the lead again – and keep it for good. I don’t have the competency to judge if that will be AGI or Superintellignece or some other variation. But, with all other computing factors changing at an exponential rate it is impossible that intelligence will not surpass the limitations of humans and human brains (including the one with a version of Neuralink).

Here’s just one data-point from Jurgen Schmidhuber: Neural networks we are using for Deep Learning at the moment have around a billion neural connections compared with around 100,000 billion in the human cortex. Computers are getting 10 times faster every 5 years, and unless that trend breaks, it will only take 25 years until we have a recurrent neural network comparable with the human brain. Just 25 years.

Hence, the yes totally.

I have a personal theory as to what happens to humans as we look out 150 – 200 years. It is not relevant to this post. But, if you are curious, please ask me next time you see me. (Or, sign up for my weekly newsletter: The Marketing < > Analytics Intersect)

AI: A conversation with a skeptic.

Surely some of you think, to put it politely, that I’m a little bit out there. Some of you’ve heard the “hype” before and are deeply skeptical (AI went through a two decade long tundra where it failed to live up to every promise, until say 2010 or so). Some of you were promised Programmatic was AI and all it did was serve crap more efficiently at scale!

I assure you, skepticism is warranted.

Mitch Joel is the Rock Star of Digital Marketing, brilliant on the topic of media, and a very sweet human being. Amongst his many platforms is a fantastic podcast called Six Pixels of Separation. Our 13th podcast together was on AI. Mitch played the role of the resident skeptic and I played the role of, well, the role you see me play here.

If you can think of a skeptical question on this topic, Mitch asked it. Give the podcast a listen…

(Play at Six Pixels of Separation)

As you’ll hear multiple times, a bunch of this is a matter of thinking differently about the worldview that we’ve brought with us thus far. I share as many examples and metaphors I could to assist you in a journey that requires you to think very differently.

If you are still skeptical about something, please express it via comments below. Within the bounds of my competency, I’ll do my best to provide related context.

Ok, ok, ok, but what about the now? (Professional)

While I look at the future with optimism (even 150 years out for humans), what I’m most excited about is what Machine Learning and Deep Learning can do for us today. There are so many things that are hard to do, opportunities we don’t even know exist, the ability to make work that sucks the life out of you easier, better, smarter, or gone.

In a recent edition of my newsletter, TMAI, I’d shared a story and a call to arms with specific recommendations of what to do now. I’ll share it with you all here with the hope that you’ll jump-start your use of Machine Learning today…

I lived in Saudi Arabia for almost three years. Working at DHL was a deeply formative professional experience. My profound love of exceptional customer service, and outrage at awful customer experiences, can be directly sourced to what I learned there.

Saudi Arabia is a country that saw massively fast modernization. In just a few years, the country went from camels to cars. (I only half-jokingly say that Saudis still ride their cars like camels – and it was scary!).

Think about it for a moment.

From camels to cars. No bicycles. No steam engines. None of the other in-betweens other parts of the world systematically went through to get to cars. They were riding camels, then they were riding cars. Consider all the implications.

We stand at just such a moment in time in the business world. You know just how immersed and obsessed I am with Artificial Intelligence and the implications on marketing and analytics. It truly is a camels to cars type moment in my humble opinion (it might even be a camels to rockets moment, but let me be conservative).

Yet, executives will often give me examples of things they are doing, and they feel satisfied that they are with it, they are doing AI. When I probe a bit, it becomes clear very quickly that all they are doing is making the camels they are riding go a little faster.

That all by itself is not a bad thing – they are certainly moving faster. The problem is they are completely missing the opportunity to get in the car (and their competitors are already in cars).

It is important to know the difference between the two – for the sake of job preservation and company survival.

Here are a handful of examples to help you truly deeply internalize the difference between these two critical strategies…

If you are moving from last-click attribution to experimenting with first-click or time-decay, this is trying to make your camel go faster. Using ML-Powered Data-Driven Attribution and connecting it with your AdWords account so that action can be taking based on DDA recommendations automatically, you are riding a car.

(More on this: Digital Attribution's Ladder of Awesomeness)

If you are moving to experimenting with every button and dial you can touch in AdWords so that you can understand how everything works and you can prove increase in conversions while narrowly focusing on a few keywords, you are making your camel go faster. Switching to ML-powered Smart Targeting, Smart Creative and Smart Bidding with company Profit as the success criteria, for every relevant keyword identified automatically by the algorithm, you are riding a car.

Staffing up your call center to wait for calls from potential customers is making your camel go faster. Creating a neural-network that analyzes all publicly available data of companies to identify which ones are going to need to raise debt, and proactively calling them to pitch your company's wonderful debt-financing services is riding a car.

Hand picking sites to show your display ads via a x by x spreadsheet that is lovingly massaged and now has new font and one more column on Viewability, is making your camel go faster. Leveraging Machine Learning to algorithmically figure out where your ad should show by analyzing over 5,000 signals in real time for Every Single Human based on human-level understanding (die cookies die!), is riding a fast car.

(To see a delightful rant on the corrosive outcomes from a Viewability obsession, and what you might be sweeping under the carpet, see TMAI #64 with the story from P&G.)

Asking your Analysts to stop puking data, sorry I mean automate reporting, and send insights by merging various data sets is making the camel go faster. Asking your Analysts to just send you just the Actions and the Business Impact from those Actions is riding a car. Asking them to shift to using ML-powered products like Analytics Intelligence in GA to identify the unknown unkonwns and connecting that to automated actions is riding a rocket.

If you are explicitly programming your chatbot with 100 different use cases and fixed paths to follow for each use case to improve customer service, that is making the camel go faster. If you take the datasets in your company around your products, problems, solutions, past successful services, your competitors products, details around your users, etc. etc. and feed it to a deep learning algorithm that can learn without explicit programming how to solve your customer's service issues, you are riding a car.

I, literally, have 25 more examples… But, you catch my drift.

I do not for one moment believe that this will be easy, or that you'll get a welcome reception when you present the answer. But, one of two extremely positive outcomes will happen:

1. You'll get permission from your management team to stop wasting time with getting the camel to go faster, and they'll empower you to do something truly worth doing for your company. Or…

2. You'll realize that this company is going to suck the life out of your career, and you'll quietly look for a new place to work where your life will be filled with meaning and material impact.

Win-Win.

Hence, be brutally honest. Audit your current cluster of priorities against the bleeding edge of possible. Then answer this question: Are you trying to make your camel go faster, or jumping on to a car?

While Machine Learning has not solved world hunger yet, and AGI is still years away, there are business-altering solutions in the market today waiting for you to use them to create a sustainable competitive advantage.

Ok, ok, ok, but what about the now? (Personal)

If this post has not caused you to freak-out a tiny bit about your professional path, then I would have failed completely. After all, how can the huge amount of change mentioned above be happening, and your job/career not be profoundly impacted?

You and I have a small handful of years when we can create a personal pivot through an active investment of our time, energy and re-thinking. If we miss this small window of opportunity, I feel that the choice will be made for us.

This blog is read by a diverse set of people in a diverse set of roles. It would be difficult to be personal in advice/possibilities for each individual.

Instead, here’s a slide I use to share a collection of distinct thought during my speaking engagements on this topic…

machines_humans_jobs_avinash

In orange is a summary of what “Machines” and humans will be optimally suited for in the near-future. (Note the for now.) Frequent high-volume tasks vs. tackling novel situations.

In green, I’m quoting Carlos Espinal. I loved how simply and beautifully he framed what I imagine when I say tackle novel situations.

Over the last 24 months, I’ve made an whole collection of conscious choices to move my professional competencies to the right of the blue line. That should give me a decade plus, maybe more if Ray is right about Cloud Accessible Intelligence. Beyond that, everything’s uncertain. :)

Summary.

I hope you noticed I ended the above paragraph with a smiley. I’m inspired by the innovation happening all around us, and how far and wide it is being applied. I am genuinely excited about the opportunities in front of us, and the problems we are going to solve for us as individuals, for our businesses, for our fellow humans and for this precious planet.

In my areas of competence, marketing, analytics, service and sales, I can say with some experience that change is already here, and much bigger change is in front of us. (I share with Mitch above how long I think Analysts, as they are today, will be around.) I hope I’ve convinced you to take advantage of it for your personal and professional glory.

(All this also has a huge implication on our children. If you have kids, or play an influencing role in the life of a child, I’d shared my thoughts here: Artificial Intelligence | Future | Kids)

The times they are a changin'.

Carpe diem!

As always, it is your turn now.

Were Collective Learning and Complete Day One Knowledge concepts you’d already considered in your analysis of AI? Are there other concepts you’ve identified? Do you think we are doomed? Is your company taking advantage of Deep Neural Networks for marketing or analytics or to draw new value from your core back-office platforms? What steps have you taken in the last year to change the trajectory of your career?

Please share your insights, action-plans, critique, and outlandish predictions for the future of humanity, :), via comments below.

Thank you.

Comments

  1. 1
    Gary Kralicek says:

    Great post as always, Avinash. The highlight to me was the point about Collective Continuous Learning.

    As this collective intelligence continues to grow, more and more complex problems can be solved for. I imagine governments and inter-country governmental alliances having the ability to solve for major issues like poverty, hunger, etc. and essentially being able to drive out the optimal use of budget dollars to best solve for the ordered set of priorities, etc.

    When thinking along these lines, it becomes really interesting if governments were to have competing priorities (i.e. world domination vs. world peace and prosperity). It makes you wonder how this technology changes the roles of governments in the future and who would hold the keys to setting and changing the priorities for which these super-intelligences solve for. Human ego and competitiveness will likely remain in that future and it will be interesting to see how this all plays out.

    I'm so happy to have received this in my inbox on Monday morning!

  2. 2

    Hi Avinash, great post.

    I thought of the collective learning and day one learning many years back when thinking through why phds now need learn more and more only to be more and more specialized. Simply because humans need to learn everything from scratch from books. Robots have no such disadvantage and contribute to a central data bank.

    Would you think humans can win if someone we can create a brain to brain interface such that human creativity can outrun robots? Or is the number of neuron connections more important in this race and we are destined to lose? Though said are'nt the brain architecture still not fully understood?

    What will robots want eventually? More of themselves?

    • 3

      Eng: I do believe that we will create some type of an interface between the human brain and the "intelligence cloud," to augment the compute power, bandwidth, knowledge etc. In the post I mentioned Neuralink – just one such initiative.

      This will spark a revolution in human intelligence and creativity for a while. I also believe that once the jump is made to Superintelligence that even an augmented human-brain can remain "competitive." I'm not sure that we would want it to.

      -Avinash.

  3. 4

    All this reinforces a conclusion someone else pointed out to me that I've been living by: Everyone who doesn't have the aptitude for getting ahead of AI in the coming decades should work toward starting their own business.

    The advice you give is well and good for probably most of your readers, but there are those of us who are entry level even in the camel world who have less hope of catching up in time.

    On the other hand, business owners will be the ones to commission AI/ML/DL *for* their business and thus be the last to get replaced. They'll be less likely to find themselves in the boxing ring staring down an AI/ML/DL robot with complete day one knowledge robotic boxing arms. The true winner at a boxing match is the owner of the stadium who sold the tickets (no brain damage required). Quick! Build a stadium!

    • 5

      I had a similar thought–that the future might involve an explosion of entrepreneurship. It seems unlikely that industries will suddenly support more competitors than they do now. But, maybe low headcount requirements of AI-powered companies frees up human talent to create many new industries. Maybe staying relevant will mean creating an AI, then charging others for making use of it.

      I also wonder if universal basic income will become more relevant in AI-powered economies. If the human mind truly becomes irrelevant outside of creating more machine minds, then it seems even entrepreneurship might become automated on day. It may be that humans simply don't have to work anymore. In this case, the massive economic surpluses produced by these economies could be redistributed among populations.

      • 6

        Ryan: I believe Cloud ML will mean that four people startups will have access to the same revolutionary tech that bigger companies are using. And, they'll have access to it cheaper. Take a quick look at the Google Cloud Platform Solutions. All the same stuff Google is using internally, available for you to use to fulfill your entrepreneurial hopes and dreams. Our peers at Amazon also have the wonderful could based Amazon Machine Learning.

        Or, consider that the software library for Machine Intelligence, TensorFlow, that powers ML innovation at Google is available to you for free.

        The second part of your comment I worry about a lot. I'm not an expert in economics or government. But, I agree with some experts who say that there will be 1.5 generations of humans who in massive groups will "useless." (Let me stress that applied here it simply means no use for in a professional context only.) The potential for social unrest, for disparity in outcomes… loads of very heavy things have to be through through. UBI might be one way to address it, I think we have to think bigger and figure this problem out.

        Avinash.

        • 7

          Avinash, thank you so much for those resources you just brought to my attention. Additionally, I really appreciate those links to the Udacity courses in this morning's TMAI. I've already started the Intro to Machine Learning course, but I've learned that I need to gain some understanding of Python, first. So, I'm taking a detour over to Code Academy or Code School or python.org to learn about that.

          It would seem the "business analytics" I learned back in Business School and along with my Ad degree is rapidly losing its utility. The "storytelling" skills are still relevant in my day-to-day, but I am absolutely seeing AI's tendrils creep steadily into my employer's marketing organization. Right now, it's automating roles we haven't filled and solving problems we haven't been able to solve, but I think that's just a buffer in between deep learning and most of our jobs. Advance warning, rather than solace.

          Being able to see the imminent transformation is a gift. We get to be scared ****less, but we also get a head start on retooling ourselves before "The Great Labor Purge of 20XX" or maybe "The Organic Intelligence Preservation Act of 20XX." Here's hoping I'm retooling in the right direction.

      • 8

        "It seems unlikely that industries will suddenly support more competitors than they do now."

        Excellent point. My thinking around this is that the vast majority of future entrepreneurs won't want to or need to build a company with a lot of market share. They simply need to build a business to replace their $50k/year or more salary. Depending on the business you create you only need a handful or few hundred loyal clients/customers to get that kind of profit.

        The way I see it, there will be wisdom in returning to the old "village blacksmith" type model where everyone in a small town had a unique skill to add value to everyone else in the town. Only now with a global customer base the niche skills will become super specialized. i.e. a law firm that specializes in the unique needs of tech startups in the food/agriculture space, or a sales consultancy firm specializing in the unique needs of marketing agencies in the wine and spirits space, etc.

        But of course as you mention, this just delays things until AIs get to the point of being entrepreneurs themselves. Then "vocation" will mean "what do you do to live with purpose and meaning?" rather than "what do you do to earn a living?"

        • 9

          Peter: This is brilliant: Then "vocation" will mean "what do you do to live with purpose and meaning?" rather than "what do you do to earn a living?"

          I believe that as we head beyond 2050 that humans will take direct and active control of human evolution. Combined with the answer to your first question, we might not look like who we are today, we might not be who we are today, we won't recognize lifespan as we think of it today. I don't know if I'm scared or excited about that, but I see the inevitability of it.

          Avinash.

  4. 10

    Yet another riveting post, Avinash!

    Complete Day One Knowledge is easily the most convincing argument for the bright future of AI. How many of us would love to travel back in time to pass our existing knowledge to the past versions of ourselves embarking on their first job after college? Beyond the professional sphere, this knowledge could have saved many of our personal relationships. :)

    I really liked your camel to cars analogy. The pace of AI/ML advancement can easily be measured by its presence in our daily lives that was not the case 5-10 years ago: ML-based pricing for our Uber rides, Facebook's face recognition ability to identify our friends in photo uploads, Gmail's spam filters and email categorization, Microsoft Word’s Fast Insider editor of your grammar and word use, Netflix movie suggestions, Amazon product recommendations, Pandora music categorization, Amazon Echo personal assistant, Google Waze route optimization, Nest learning thermostats.

    Our natural thinking is that even if algorithms can predict the future results and suggest the best plan of action; ultimate decision is made by humans who can see the gray matter beyond the binary limitations of intelligent machines. Interestingly enough, we don't always want to carry the weight of such decision-making power and would gladly outsource this task. A well-documented Freakonomics experiment found that people would rather flip the coin on life altering decisions including quitting their jobs and ending their relationships:
    http://nymag.com/scienceofus/2016/08/a-study-got-people-to-make-big-decisions-with-a-coin-toss.html
    One practical example would be for AI to help the interviewing panel make the "right" choice between job candidates in the final round of interviews, by far not an easy decision to make.

    AI critics can easily see the frightening military potential behind advanced robots developed by Boston Dynamics:
    https://www.ted.com/talks/marc_raibert_meet_spot_the_robot_dog_that_can_run_hop_and_open_doors
    After all, if various WMDs are not precise enough, would the future wars utilize Terminator-like army, far more advanced than the human one?

    On a brighter note, most of the leading tech giants not only recognize the potential behind ML, but also democratize it to potentially expand their reach: Google offers TensorFlow as open-source software; IBM Watson engine has a free subscription tier, making it easy to turn your Raspberry Pi into a talking robot; and one can code their own voice-based Alexa skills using Amazon Web Services.

    • 11

      Alex: I am sure there will be a day when AI will save our personal relationships! (This is not a joke: AI to give love advice to troubled hearts. Oshi-el is amazing, even if not perfect yet.)

      Deep Learning is already being applied to employee related decisions (though the service is not yet available in the cloud for all of us use).

      I am a software person and not a hardware person and hence I don't have the competency to share anything of value re where hardware will progress. It feels like many permutations exist, for our sake I hope the software will choose the ones that are good for humanity for the hardware.

      Avinash.

  5. 12

    Thought and info rich post and good commentary. Thank you Avanish and fellow readers.

    When thinking of AI and how some fear it'll make we humans irrelevant, ask people "Do you believe that everything that should be done today, is already being done?" Think of how we pollute our oceans and our air. Might we learn how to explore our solar system even faster? Can we enhance the quality of life on Earth for not just humans, but all living things and can we do so more sustainably? These are the sorts of challenges I hope we can augment ourselves with by using AI. It will always be computationally more powerful than us (because we designed it to be so), but it'll need training from us to learn what sorts of problems it should solve.

    And if we succeed in teaching it that, it will be nothing less than the next evolution of consciousness.

    • 13

      Aaron: Bravo!

      Of all the things I'm excited about in context of intelligence, the one I'm most excited about is that it will figure out how to solve intractable problems like Climate Change (a problem for which we are already past the point of no return, sadly).

      Another thing to add to the how should we think about this hopper… I believe we will take control of evolution (atleast of humans), and that presents possibilities most people don't think about when it comes to our future.

      You used the word consciousness. It is one that I'm uncertain about. Mr. Harari defines intelligence as the ability to solve problems and consciousness is the ability to feel things. I'm uncertain if we should focus on consciousness at all. Its a big topic!

      Avinash.

  6. 14
    John Davidson says:

    I'm blown away by how consistently you manage to find connections between disparate threads to identify something profound. The first part of this post is the latest example.

    Like many others I feel I am keeping abreast of the implications technology presents. The two conclusions you drew still surprised me. I'll come back to comment with something insightful. At the moment I'm a little too scared even as I am inspired.

  7. 15
    Vivek Deshmukh says:

    You are rock solid Avi, and we find something new every time we read your new posts.

    But one thing I am still trying to convince myself (honestly concerning) is the ability for machines to take over creative jobs. And I don't mean creative as in creating something new out of the 'data' that is out there but the ability to makes sense out of nothing and provide a new dimension to our thinking. E.g. can a machine look at a falling apple and discover gravity?

    • 16
      Trotte Boman says:

      A falling apple is data. Realizing there is some underlying force that drives the apple to the ground is exactly the type of things an AI could learn better than a human.

      However, the machine would simply learn that "after a while (or in the computers way of thinking, according to some distribution of time) an apple falls from the tree with a certain speed". It does not reflect on WHY, it simply learns that it is so.

      The question "Why?" that Newton asked himself when he saw the apple fall is still a question only humans care about answering though

      • 17

        Trotte: An excellent point! Thank you for making it.

        If you checkout the article I link to in the New Yorker, the journalist makes this point as well. The skin doctor is already outmatched on the What, and that will become pervasive across fields. The skin doctor's real value at the moment is Why (as in, why did the patient get skin cancer). Her ability to get the Why comes from experience, from being able to connect dots across patients the AI does not have access to, and, her ability to really listen.

        At the moment, the AI can't do any of that. (You can see how this is connected to what I say in the green text on the slide in the post.)

        I am a data person. As I read the article, the doctor's fantastic ability to solve the Why is a data problem. We don't have the data she is using to ask the questions, we don't know how she is connecting the points. But at some point we can get that data, say brains can be read like your USB Flash Drive or by simply asking 10,000 doctors to document what they are thinking, then a deep neural network can learn.

        Think even beyond asking the questions. Consider the data collection mechanisms I mention in my reply to Jorge. If we have these new sources of data (nano-sensors etc.), then the deep neural network can preemptively identify and resolve.

        Until then, we humans own the WHY! :)

        Avinash.

  8. 18

    Love the article.

    One quick 'fix': "Bonus 2: While are on the subject of objects… ", should be "While we are"

  9. 20

    Hi Avinash,

    It is one of the brilliant post which has blown away many confusions about ML and AI. For me, it is really from Camel to Car moment. Beautiful part of the article is that it is comprehensive.

    I took seriously your every suggestion. Following your blog since2012.

    Last month I had convinced to key decision makers about the SEE-THINK-DO-CARE framework and last week I've seen this post. (I missed this post in Newsletter)

    Here few questions:

    1. What will happen to people who are still working in Non-IT industries like textile, real estate, teaching, agriculture, HR, manufacturing, CNC related jobs, architecture, Civil….

    2.There would huge unemployment problems in India. Still, we are lacking innovative blue collar companies like Sillicon valley.

    Can you please jot down wonderful guide for people like us who really want to some wise guidance, especially people who are earning their bread in Digital Marketing Industry and are coming from other backgrounds like Commerce and Science. I have big hope from Market Motive here.

    3.Your post has reminded me of "The Atlantis" from "Atlas Shrugged" by Ayn Rand. The only difference is machines are becoming prime movers and humans became second handers. Even today we are disthroning people like Vishal Sikka who was trying to demonstrate that elephant can dance.

    By Considering the population of Indian youths there would be limited jobs and the huge pool of people who will try to enter and hence bottleneck.

    What professions/businesses those people shall adopt who will not get the job for these new technologies (AI/ML). People with the skill level of scale 5-6 and not 7 and above. They will be jobless.

    I am sure you've got what I am trying to ask.

    Your answer would be a great help.

    Thanks again and thanks a lot.

  10. 21
    Lindsay Sterne says:

    People tend to think of AI from a hollywood sci-fi perspective. You've masterfully illustrated that AI is so much more than what the movies have managed to conjure up.

    It was great to have your slide at the every end. It has caused me to think about what I need to do to join you on the other side of the blue line. I also enjoyed hearing the podcast and your thoughts on some of the scary stuff that people constantly bring up with respect to AI.

    Thank you for staying ahead of the rest of us and making the bleeding edge a lot more digestible.

  11. 22
    Jorge Nogueira says:

    Hi Avinash,

    One brief question: if you write COMPLETE FIRST DAY KNOWLEDGE you refer to digitized knowledge, only, right?

    How does non digital or gated information fit into this?

    • 23

      Jorge: No, I don't mean digitized knowledge as we think of it today.

      There is data about your website, the flight patterns of aircrafts, real-time weather patterns, shifts in rates of poverty in the world, etc. that is currently available in a digital format. With every day, what's available explodes.

      There is data we don't have, but will become available over time.

      I don't have nano-sensor in my body that are constantly collecting data about me and feeding it to a Health AI. This will happen soon enough. It will be available digitally.

      We don't have data about how I behave if the temperature is different or have a overloaded calendar or a week off work or meeting new people or when have music playing or… A lot of this data will be collected and connected together with some Prediction AI (that among other things can make my life better). We need the sensors that will do all this, we will need them to be sophisticated beyond what we can imagine today. But, it will happen.

      Hence, I don't think Compete First Day Knowledge will be limited to what we today think of as "digital data."

      Avinash.

  12. 24

    Hi Avinash,

    First of all, let me vote for your experiment of starting TMAI. I find it very useful and look forward to receiving and reading the same.

    Artificial Intelligence with the co-relation of Collective Continuous Learning + Complete Day One Knowledge changes the game completely, which is hard to see favourable of human intellect as we speak.

    I totally agree to the point you are making with assumptions of Tackling novel situations, that the only thing which is currently at play and let us charge a premium or a sufficient profit margin to our services. In absence of which we are no more than a commodity which is replaceable by other humans, right now. And now I see I am also competing against the machines powered with AI.

    Future did look cloudy to my intellect, and I am worried. I understand there are no easy solutions to compete against the given power of Collective Continuous Learning + Complete Day One Knowledge.

    So what if we do not compete, what if we start exploring avenues and areas where the human contribution becomes indispensable. Are there any such areas that exist? Is there a possibility of such roles only doable by humans?

    I am very interested to learn from your understanding – what are the future professions, or means of livelihood. I think if we start on the journey of tackling novel situations, we might be able to strech the capacities of our brains further and can come up with better things to do with an active support of AI. A NEWER kind of evolution age, which is not in competition with AI.

    It is a fanciful thought! Fanciful because we are still stuck in our past, and the journey to enhance has not really started as an effort for masses, but those who understand the complications, and make use of the time NOW – will surely be in the position of advantage – the new rich class of the future.

    • 25

      Gaurav: There is a lot in your post, allow me to pick up one of the threads.

      It is important that we have to start now, and figure out our individual path to tackling novel situations. They exist in every profession because every job has "jobs where discerning the fine line between good data from bad data is critical." In HR, in Customer Service, in R&D, in Design, and more.

      At the moment it is unclear, looking 50 years out, if there will be job/areas that will be exclusively the domain of humans. Looking a century out, I feel certain there will be no such things a exclusively human.

      Over the next handful of decades, if we solve Neuralink or some variation of it, what will be exciting is to have AI-Augmented-Humans. That will allow us to solve tasks of currently unimaginable complexity, creativity, and and more. In a sense, it will be a period where we'll be the very best humans that humans could ever be.

      You can see a very small manifestation of this already. Both Ke Jie and Lee Sedol are better players after having been defeated by AlphaGo. The algorithm's play helped them imagine new possibilities, new creative moves that had never been thought of in the multi-thousand year history of Go.

      Avinash.

  13. 26
    Ironprincess says:

    All that you say is true…I really do not think people fully grasp the power of AI which is why we keep using it for the wrong things. The real value of AI isn't in what I call "two-trick ponies," using AI to just do the same wrong problem solving activities we've been doing for years. Your example of the "car" using AI to target people who are ripe for a financing offer, that's like using an AR-15 to shoot a deer overkill.

    I wish we could understand AI so that we can think bigger. Right now the homogeneity of this discipline is putting us on a dangerous path of using technology in a limited capacity and not solving actual problems that need to be solved.

    I'll be optimistic about AI when we diversify its use, democratize it's understanding and start really tapping into the power of rapid, fast learning to eradicate some of the world's most vexing problems. As long as it remains a tool of capitalists and non-human centered designers we're all doomed.

    Very few data scientists and AI researcher start with people and problem first, technology.

    That's a sweeping generalization but it's a disturbing trend I've seen in academia and in professional world. Many of the algorithmic modeling used in AI is rife with bias and discriminatory data sets.

    Before we can run we must walk. And that means we must retroactively fix many of the faulty premises, findings and research some AI implementation has been founded upon, ESPECIALLY in marketing.

    Otherwise it'll just become another oppression tool of the have to keep down the have nots.

    • 27

      Ironprincess: I feel like we are very early in the evolutionary process, and the proverbial intelligence explosion awaits around the corner!

      Consider though the examples I've mentioned in this post about Skin Cancer and Diabetic Retinopathy. That solving a material problem for a billion humans. There are loads of revolutions happening with health, all would fit your definition of using AI for the "right things."

      You are correct to see the limited scope that a number of Scientists, Academics and companies are setting for their AI initiatives. It will be nice if we make smarter chatbots to do tech support for a laptop manufacturer. I do hope people who are investing in that start with a more audacious goal, and then settle on the finite chatbot.

      Avinash.

  14. 28
    Girish Pai says:

    Most of our economy related issues should get solved with AI coming into the picture. main concern is how are we going to use this technology. is it in the right direction or not. we as consumers are already getting bot calls on a daily basis, whether to say promote products or to harass people, a line has to be drawn. with deep learning stock markets get predictable but at one point the market is going to behave differently and the technology becomes useless, unless it can predict 100 percent, which is an unlikely scenario. This post is awesome as always and will be waiting to read more

  15. 29

    Hey Avinash,

    Great article as always. I do not (yet!) have a firm technical background in the workings of AI. However, I'm wondering how to maintain a standard of ethics in an AI future.

    These machines are great at optimization but if they get so complex that nobody can understand how they work, how will ethical overwatch be provided to make sure these tools don't harm peoples' lives?

    Cheers,
    LH

    • 30

      Lazare: Ethics is the trillion-dollar question. I'm afraid I'm not an expert in the area of governance hence I don't have specific guidance to offer except say that it is important.

      If you want to imagine how important, an algorithm is replacing bail hearings in New Jersey (and ditto in the UK as to which criminals get bail). If you've read anything about Deep Neural Networks, you'll quickly learn that they are wonderful and getting better every day… But we really don't know how they learn. Consider that and the application of an algorithm to the criminal justice system. (Though, in the US criminal justice is so broken today, see an org I support, The Marshall Project, that the AI can't possibly do worse.)

      I'm heartened by gatherings like the Beneficial AI Conference (more in TMAI #88) where the world's bleeding edge thinkers are coming together to create structures and guide AI development.

      Avinash.

  16. 31

    I work in a field that already has some AI involvement (estate planning and elder law) but to be highly effective there is a level of personal customization required and it's difficult to imagine how AI may be able to step into that space, however, I'm sure it's not out of the question.

    • 32

      Liz: There is certainly a ton of complexity to the legal profession that might take us some time to get through.

      Some of the regular, perhaps more painful, parts of the legal profession might benefit sooner. I'm thinking of solutions like Kira who provide machine learning contract analysis. Very cool. I'd mentioned the influence on the bail process in a comment to Lazare as well.

      From a broad legal profession perspective, this article has valuable coverage: A.I. Is Doing Legal Work. But It Won’t Replace Lawyers, Yet. If everything AI can do today is put into place, it would save 13% of a lawyer's time.

      The impact on the broader work done by Attorneys in the criminal system, the ability to connect the dots, solving puzzles, arguing in courts might have to wait a while.

      Who knows, if we get to a Minority Report type future – and honestly there will be enough data available that we could – then maybe we won't need lawyers at all! :) Ok. Kidding.

  17. 33
    Florentino says:

    Thank you Avinash!

    I totally agree with you. My only concern is How to implement AI or ML.

    Hope you can give us some ideas.

  18. 35
    Mike Ricker says:

    Hi Avinash,

    Great post! I am very interested in the Adwords component of your post with respect to moving to smart targeting. Are there product features in Adwords that one can turn on to take advantage of ML? Are there any resources you may recommend for someone that wants to take a deeper look at this from a small business point of view?

    Thanks!

  19. 38

    Avinash-

    Your articles always are an inspiration and cause thought. The potential of good that could be brought about is simply amazing. It makes one want to be a part of it. Yet there are many potential road blocks and areas where that kind of intelligence is of high concern.

    The medical portion is amazing and it would be wonderful to actually be able to be diagnosed with conditions at early onset and not have to chase after 20 doctors to find out. To potentially save many more lives. That thought just makes my heart feel good. While I believe the intelligence is there I, unfortunately do not see our government,doctors or insurance companies permitting such a thing. The money involved in making us bounce around to different doctors for each different portion of an illness is unbelievable. That money effects the economy. The only way to recoup the losses that would be incurred would be to make the AI testing astronomically priced. Insurance would most likely require we jump through 20 other hoops before being considered for AI testing and even then probably wouldn't cover it. (Like they already do)

    A comment on here about government use made me think. I have no doubt they already have several forms of AI but the better it gets the more then can do and honestly that scares me. I can't imagine giving some of our world leaders that kind of power.

    When AI has more knowledge than humans, who then controls AI??

    Keep doing what you do! You are an intelligent, insightful, thought provoking person.. I look forward to your posts every week.

    • 39

      Spring: This should not matter, but let me assure you that the incentives are there for the insurance companies to support AI-powered healthcare advances.

      Here's the reason: A billion dollars of cost is slated to be saved for the UK NHS from the focus on AKI (Acute Kidney Injury). In the US $38 billion in costs, yes, thirty-eight, can be removed from optimizing detection of Stroke.

      Of course, saved lives should be enough.

      The last part of your comment keeps me awake worried. I sadly don't have the expertise to give you any good answers.

      Avinash.

  20. 40

    Step right up and board the roller coaster of Artificial Intelligence!

    You don’t want to miss this one folks. It’s been through the cycle twice before but this time it’s different. There will be no chasms of disappointment or other scary drops because we now have even more amazing demos!

    Yes, that’s right—demos!!!

    Over here are robotic arms that can select objects from a pan, and they learned how to do this all by themselves. Wow!
    Never mind that Terry Winograd had robots picking out objects in the last AI Bubble where the bottom fell out, this is new, New, NEW!

    The demos have always been great for AI. Even the old AI from Bubbles Burst in Bygone Days we had:

    – Medical diagnosis better than what human doctors could do. See Mycin for prescribing antibiotics, for example.
    – All manner of vision and manipulation. Blocks? So what. Driving cars? Yeah right. Turn ‘em loose against a New York cabbie and we’ll see how they do. The challenge for autonomous vehicles has always been the people, not the terrain. No matter how many autonomous cars drive across the dessert (talk about the easiest possible terrain), they’re nowhere until they can deal with stupid carbon units, i.e. People, without killing them or creating liability through property damage.
    – Math. Computers have been solving mathematical theorems for ages. In some cases they even generate better proofs than the humans. Cool. But if they’re so good, why haven’t they already pushed mathematics ahead by centuries? Something is not quite right with a demo that can only solve theorems already solved and little else.
    – Games. Oooh, yeah, computers are beating chess masters! Sure, but not in any way that remotely resembles how people play chess. They are simply able to consider more positions. That and the fact that their style of play is just odd and offputing to humans is why they win. What good is it? What are those algorithms doing to genuinely add 1 billion to IBM’s bottom line?

    There’ve been a lot of great AI demos over the years. There’ve been a lot of very smart people, experts in the field of AI and not just gushing reporters and futurists, who’ve proclaimed it’s right around the corner.

    Most all of them have claimed it’s just a matter of computer power. And we keep getting more and more of that. But it’s far from clear that AI has really gotten that much smarter.

    Avanish, you say we just need 100,000 times faster computers. Woo hoo!

    And many are predicting exactly when the singularity will happen based on that. Computers get 10 times faster every 4 years so it’ll only take 25 years to get computers 100,000 times faster.

    Whoa there big fella! Moore’s law actually says they get twice as fast every 2 years. So after 6 years, they are 8 times faster, not 10x in 5 years.

    It’s going to take more like 34 years to get there. And there’s a LOT of assumptions behind that:

    What if Moore’s Law doesn’t hold?

    It hasn’t been easy keeping up, and nobody thinks it will go on forever. Yet we need it to continue another 34 years without slowing down just to have a shot at this AI thing.

    BTW, just being able to build a computer powerful enough that takes a giant data center to house doesn’t mean it will make sense to use that computer to take over minimum wage jobs. It’s an expensive piece of machinery for many more years. We can build one in 34 years and then how many more years until we can fit one cheaply on a desktop? 10 years? 20 years?
    What if neurons in the brain are significantly more powerful than we think?

    There is credible evidence, for example, to suggest that they may be capable of quantum computing on some level. That is going to change the math of getting to the power of a human mind quite a lot. We’re way behind at creating quantum computers on that scale.

    What about the software?

    The first 2 AI hype cycles failed because we had chosen essentially the wrong architectures. First it was logic and exhaustive search, then it was expert systems and knowledge representation. Now we think it is neural networks, and because they can learn, we’ll just let them learn and not worry too much about software.

    But there’s a whole lot more that we know about humans and animals now that isn’t addressed by that model. How are instincts programmed in? What’s the role of emotion and hormones and how are they implemented?
    Think of every hard question having to do with why human minds do what they do. Think of how intractable mental illness is simply because we are largely clueless.

    Isn’t it hubris thinking we just need more powerful neural nets and we are there? You bet it is.

    Artificial Cockroaches

    So we need artificial brains that are 100,000 times more powerful. In essence, we can compare today’s AI to brains the size of what cockroaches have. Yet, we’re worried they’re going to take all of our jobs.

    Are you in a job that a cockroach could do? I hope not. So far, I am not aware of anyone having harnessed cockroaches to do their bidding, but they are cheap, plentiful, and just as smart as today’s AI’s. Maybe smarter if their brains are quantum computers too.

    Maybe it would be cheaper to spend billions learning how to make cockroaches useful?

    I don’t know, but we don’t even seem to be able to make much smarter animals useful. Are there dogs running machinery somewhere in China? Is a particularly adept German Shepherd behind the latest quant trading engine on Wall Street?

    Nope.

    Get ready for the trough of disillusionment for AI, people. It isn’t going to take your jobs away. In fact, quite the opposite will happen. Just as has happened in previous AI bubbles, Ai jobs are going to go away.

    That’s how the third AI Bubble will end. But take heart, AI aficionados. There will inevitably be a fourth AI Bubble. It will happen before the 34 years brings us powerful enough machines, but it will be close to that epoch. It will burst too, because we’ll blow right through the creation of 100,000x more powerful machines and be left wondering why they didn’t work.

    Why won’t the machines talk to us?

    My theory about that, BTW, is the machines are going to be much like Helen Keller. She was an extremely bright person. Yet because she was blind and deaf, she couldn’t communicate for a long time.

    People are not very patient. We will eventually create artificially intelligent machines, but at least a few generations of them will be killed before we realize they really did work simply because they didn’t communicate well enough to convince us in time.

    BTW, people have been doing collective learning for a long time. That’s the purpose of language, reading, writing, and civilization in general.

    • 41

      Bob: Great progress will never come without great skepticism – the people trying to push the world forward need motivation. :)

      Enough has been already delivered, and you are using it already, to know that practical revolutionary use is already happening. If you've used Google Translate at all, your demo theory is imprecise because Translate is used by millions every day and is completely running off a Deep Neural Network. (Oh, and it just learned how to translate how to translate from Russian directly to Japanese without any human involvement!) Deeper technical article here.

      I offer two more thoughts for your consideration.

      Chess is brute for computation. Go has more moves than, Bob, stay with me here, the sum of atoms in the entire known universe. AlphaGo beat the very best human players not by brute force computation but by mastering intuition. (Technically a combination of a Policy Network and a Value Network. Deep technical details here.)

      Please don't limit your imagination related to hardware and compute power to traditional strategies. Google invented Tensor Processing Unit to do something revolutionary: Train AND execute neural networks. The hardware, specialized for ML, delivered 15–30X higher performance and 30–80X higher performance-per-watt than contemporary CPUs and GPUs. An unimaginable leap. Deep technical details here. And this was the company's first attempt at this.

      You are right about Quantum Computing. I did not discuss it because it is very early times for now. Yet, mind-blowing stuff to come. (There is speculation that a quantum computer that can perform a fiendishly difficult task that no classical computer is less than 12 months away. With 50 qubits!)

      Let's put something sweet where your words are coming from. If in five years you don't agree that atleast 80% of what you say in your comment is demonstrably imprecise, I'll buy you a cup of coffee. If for some reason we can't meet in person, I'll send you a gift card for two coffees (mine and yours). Game? :)

      Avinash.

      • 42

        Avinash, thank you for your thoughtful response!

        My short answer is I will take your bet! My long answer is posted on my blog as it includes pictures:

        http://bobwarfield.com/avinash-kaushik-responds-ai-cockroaches-post/

        I will tell you by way of a tease that Google Translate and your response illustrate my points well.

        If I enter the first two paragraphs of your response into Google Translate as English and ask it to translate to Russian, it happily does so. If I then ask it to translate back to English, it also gives me a lovely translation:

        "Great progress will never come without much skepticism – people who are trying to advance the world need motivation. :)

        It has already been delivered enough, and you are already using it to know that practical revolutionary use is already taking place. If you used Google Translate at all, your demo theory is inaccurate, because Translate is used by millions every day and completely disconnected from the Deep Neural Network. (Oh, and he just learned to translate from Russian into Japanese without human intervention!)"

        Just one problem–it's seriously wrong in a way a human translator never would be. It says Translate is "completely disconnected from the Deep Neural Network" instead of your original meaning when you said "completely running off a Deep Neural Network".

        There's a lot more, including the reasons why if the human brain is a quantum computer, it will take us 70 years not 30 to get a quantum computer powerful enough to compete.

        Cheers!

        BW

  21. 43

    Avinash:

    I am a total fan of you, I have taken all your classes at Simplilearn and read most of your posts. I love how you can break down complex matters into simple terms, but most of all how you always point out the "so what" in the issues you address.

    In this case, I think that one obvious "so what" has to do with social constructions and economic systems. Automation and AI are here, we are moving forward with huge improvements to human life standards, but if we don't consider universal income seriously, and re-think what capitalism and wealth means, we will be at risk to have two parallel universes; the ones that own technology and means (brains and money) to operate it and the ones that will be left behind with no use in an economy in which their physical or less sophisticated mental contributions will be minimal.

    Forget coal or automotive workers, most of the workers won't be needed, and I am not talking just blue collar workers. most of the trades as we know them won't be needed. Very few could be re-trained, there will be many fields in which machines will be better than humans. Could you re-train camels to compete with cars?

    What's your take on this?

    • 44

      Adriana: This is an important matter for our governments and policy influencers to think about carefully, now (!), and start to have the discussion so that new programs and policies can be put in place in the next couple of years.

      I touched on this in my comment to Ryan as well. I agree that there will be a generation and half of humans that will be "useless" (as in no use for in a professional context). This will harm developing nations and lower economic/educational segments the toughest (so the people who need most help even today). Universal Basic Income might be just one of many, many more solutions that need to be explored.

      Avinash.

  22. 45
    Ana Mourao says:

    Hey Avinash,

    I would like to (humbly) suggest another book on AI: Heartificial Intelligence: Embracing Our Humanity to Maximize Machines by John C Haven.

    The main argument of this book is that we (humans) must really have a deeper understanding of what it really means to be human so we can better program AI moving forward.

    The book may not be that technical but it is a very interesting point of view that I could (not yet) find on other books about this subject.

  23. 46

    Hi Avinash,

    No, I don't think you are "just too deep into this stuff" as you put it.

    It's real and it's happening fast and you are right that we are not prepared. I'm with you on the last AI | Humans slide. For me, it's too volatile to even worry about all the possibilities and just focus on building my business and brand to (hopefully) better weather the inevitable storm. Please keep us updated on your thoughts on this topic.

    Thank you,

    Alan

  24. 47

    Avinash,

    This post (and your previous post on AI) has stoked a lot of thoughts in me. I’ve been in digital marketing for the last 10 years, and have worked in media agencies, Facebook, and Google. I’m a smart guy and have always been interested in self-improvement.

    Pockets of smart people I admire are stating they cannot overstate the impact that Artificial Intelligence will have, and I’m listening in earnest. However, I feel like my skills and the trajectory I’ve been on will be swept away without investing some time and energy into learning some of the more logic-driven disciplines like maths, physics, and computer science.

    I have shied away from these subjects since high school and feel like now is the time to catch-up before I get even more left behind. I’m going to start with that Udacity course you’ve mentioned on your newsletter and if I can’t wrap my head around that, start simpler until I get to a point where I start absorbing.

    Thanks for the post, I feel like we’ll look back years from now and wonder why weren’t more people excited!

    Thanks,
    Luke

  25. 48
    Vanessa says:

    Sorry to be off-topic, but due to my tracking OCD I must inform you that you added newsletter UTM params in your link to your previous post (Artificial Intelligence: Implications On Marketing, Analytics, And You).

    • 49

      Vanessa: I wish you can see how big my smile is after reading your comment.

      I should have expected that my brilliant audience will not only read every word I write, but that you will click all three instances of that link AND have display big enough to notice the tracking parameters AND know that that is wrong AND tell me.

      You are amazing. I'm glad you have OCD. Thank you. I've fixed the code.

      Much love.

      Avinash.

  26. 50
    Jeremy Lavitt says:

    I thought an interesting path to go down would be law/trials. Knowing all precedence and applicable laws, success of penalties couldn't AI help determine a more equitable judicial system. We could help reduce biases caused by being tired, sloppy lawyers, racial bias etc.

    However, how would AI be able to understand extenuating circumstances? Show mercy or compassion? Maybe initially it could be a check against bias allowing for different sentencing, retrials, grading of judges?

    How could it change elections? Could it write a paper guaranteeing an A while avoiding plagiarism? Shouldn't we be able to fully predict the weather?

  27. 52

    Thank you, Avinash!

    You mention "ML-powered products like Analytics Intelligence in GA" – is this what is now called "Custom Alerts" in GA? I can't find anything called "Intelligence" in GA, nor in any articles newer than 3 yrs old. And I can't find it in Discover either.

    Just want to make sure I'm not missing something.

    • 53

      Kirk: Good question, I should have clarified.

      Open the Google Analytics Mobile app, on the top right you'll see a a little circular icon, click!

      If you have the Premium version of GA, on the top right you'll also see the same icon, but also the word Intelligence. Click!

      Avinash.

  28. 54

    On point as usual Avinash. Thank you for your thoughtfulness with this stuff.

    A question for you: I've been in tech for 20 years, mostly as a designer and product manager. And, I want a change and am looking to acquire a small business. One is landscape design/build and the other is a commercial painting business. Both are very profitable and have been for years. How would you think about businesses like this in 2017? Painting is labour intensive, often repetitive work. Do you see a lot of that labour being replaced by robots in the next 10 years? If so, does that not suggest that whichever painting companies are brave enough to leverage AI to improve labour efficiencies early, and the most likely to win?

    Very keen to get your thoughts on this example.

    • 55

      Adam: There is little question that automation will accelerate, and that unlike in the past the automatons (robots) we create will have smarter brains capable of learning – simple things, then simpler complex things and then complex complex things.

      I'd mentioned in the post that I'm not an expert on the hardware side of things, I'm spending all my time on the software side (brains). To the extent that I can stand outside and look in, we might not have the automation of commercial painting in the next decade. Both robotics need to make a lot more progress and because I don't feel like it is a commercially attractive – at this early stage – problem to solve.

      Here's a report I was reading from Citi, published in Aug 2017. It specifically covers robotics and focuses on the retail sector. It has ideas and implications that would be of value for you as well.

      ~ Technology at work v3.0: Automating e-Commerce form Click to Pick to Door (pdf)

      Avinash.

  29. 56
    Sandeep Goswami says:

    This is a great follow up content to your previous content about Artificial Intelligence.

    Having said that both shared some very good insights. I have been keenly following your blogs and would recommend them to everyone who wants to be updated on all kinds of digital mediums.

    Thanks – Keep coming with more such posts.

  30. 57
    Eshika Roy says:

    Thank you for posting this informative blog on the possibilities of artificial intelligence.

    AI is developing at a breakneck speed and is continuing to impact every aspect of our lives. I can't wait to witness the fruition of the potential of AI and what it holds for us.

    Looking forward to your next post.

  31. 58
    Laura Artibello says:

    Intrigued? I am.

    Full disclosure, did not read every word or listen to podcast. Get the gist, as we humans used to say.. New to all this learning from the young, bright, disciplined & excited AI students & young workers emerging. My mantra is learning new each day. Evolution of AI is here. Each generation builds more, faster, smarter. We are obsessed with new stuff!
    Driverless vehicles were once laughed at; AI is not nor should be ignored.

    What has not changed remains (miss) understanding because, the whole damn thing is so complex & confusing. Will anyone need to work? Who will feed the children? Will robots get divorced? Do I still get my pension?

    Quoting an AI entrepreneur, we still need 'the human in the loop' and I think we can agree, we're a long way off from that? Extreme shift each year? Yes as I've seen in my nearing 60 lifetime and my granddaughter will know no other. Recall; it was us BB's that wanted telephones in our cars!

    Collective Continuous Learning. Can we as humans not adapt to this? We are teaching our 'bots' to be precise & skilled assistants. So why not instruct the human mind too? Can we play together?

    World population continues to grow. People build robots, systems, programs. Human value still needed. Where we replace humans with AI bots, systems & programs is for us as a responsible world society will be to distinguish where best to develop the indestructible forces? Safe lives from natural catastrophe? Move food safely through war torn countries to the needy?

    I believe we should find what is not possible before we develop the plan to deep. And nothing is impossible; really. If we are building bots to chop vegetables, drive buses and order in Chinese food for us we're playing with extreme laziness, little learning let alone collective continuous learning. AI will do so much more.

    Exciting times to be in science, engineering, creative & monetary driven environments.

    Forward – always forward. Thank you Mr. Kaushik, you interest me.

    AI

    • 59

      Laura: Over a long period of time, humans will not have to work, all humans (including kids) will be well fed, and pension as a concept might not exist because your core needs will be addressed regardless.

      I do not believe that humans need to be in the loop. In fact, the very core of Deep Learning is that we don't even understand how these algorithms learn – except at an abstract level. As we approach and move past Superintelligence it will be like a squirrel trying to be in the loop of how humans think or behave.

      So what the heck do humans do? In terms of the near future, I touch on this at the end of my post. I also want to underline that new jobs will get created that we can't even imagine right now. (Small example: Would you have believed anyone who said just 10 years ago that people will be able to make videos about themselves and make a living on YouTube? I mean, what!) And, I am optimistic that we will figure out how to give humans what they really want (really, really, want) and let Intelligence (artificial, general, super) direct rest of the strategy of this planet.

      Avinash.

  32. 60
    Peter Byre says:

    Amazingly conceived article as always and I will certainly need to re-read to fully grasp all of the themes within.

    I hope this is on subject…A search on the page for the word "trust" didn't reveal anything. It is hard to deny the lack of suitability the human brain has for the range of tasks you describe, but how do you think the trust issue will be overcome.

    You (and I am guessing a large proportion of your readers) have a high exposure to technology, innovation insight (etc etc) – how about the millions that don't, how will they become excited and accepting, rather than scared and dis-trusting in the relatively narrow time-frames that you quote.

    An exciting future no doubt!

    • 61

      Peter: I excluded many things from the discussion, one of it was Trust.

      Primarily because it is a complex topic, it does not deserves a drive-by treatment. Let me try to summarize two facets of it.

      One critical thing I've had to be comfortable with is in context of Trust is that humans can't really understand how Deep Neural Networks learn. Not that we don't want to, we simply can't because of the complexity.

      This is hard to let go. We as humans are programmed to want to know who things work. In my meeting this morning I saw a client who refused to use ML for their advertising as they don't understand how it works (even though their revenues are down for seven straight quarters and ML can be a part of turning that around!).

      At some level, I've had to be comfortable that if the output of these narrow AI systems is what we want, it is ok not to understand exactly how they improve clicks or profit.

      On a grander scale, I'm afraid I'm not qualified to answer. Who is going to make sure the AI's that get built are solving for a commonly accepted collection of guidelines? How do we build checks, can we even build them into our algorithms? What role does government have? I don't know. I am excited though that the smartest people in the world are getting together and thinking this through. More here: Beneficial AI 2017

      I do deeply wish that senior levels in the government were more aware of the massive change in a short time and encourage more research on AI safety, encourage development of new programs for citizens to get trained, even new models of getting paid, so on and so forth.

      I hope this helps.

      Avinash.

  33. 63
    Alan Potts says:

    Great blog on artificial intelligence. I loved reading it.

    Eye catching point for me is the Collective continuous learning.

  34. 64

    Hey Team

    I just wanted to point that video -> Human Need Not Apply

    https://www.youtube.com/watch?v=7Pq-S557XQU

    : )

    • 65

      Dim: It is a thought-provoking video, and so much has already changed since 2014!

      I find it easier to split hardware from software when I think about the future. It is one small flaw I would like to point out in the video, it is very hard to truly grasp the implications. For example, from the video it is very hard to get to the conclusion that thanks to the software implications (machine intelligence) the jobs that'll go first are white collar and not the blue collar ones that we normally think.

      Over time, as progress is made with hardware AND software, the implications are truly mind-boggling.

      Avinash.

  35. 66

    Hi Avinash,

    Long time no speak!

    Thought provoking post. I had thought about the knowledge from day one idea and that's why I'm taking the steps to get into this now. I don't know whether what we're going to do is right but it's now possible to try with all the tech available to us at such low cost today, 5 years ago it was expensive. Now it isn't. I hadn't thought about collective learning but of course that makes total sense.

    My question is related to the knowledge from day 1 idea in our field. I've seen some open source algorithms we can start to use to train on our own datasets but which in your opinion are the best ones to start with? Is there a future post in this?

    Thanks as always.
    Steve

  36. 68
    Negar Zandipour says:

    OMG Avinash, what a great and still overwhelming article to digest. I read the whole article and it is interesting to start thinking of how AGI can be used to solve unknown unknowns

    Thank you for that

  37. 69

    Love it Avinash….have been following you for a while now.

    AI seems to be the future. Even with apple new iphone launch we can see the new technology integration.

    Also, in my industries like search AI has been a huge part of Google's algorithm.

    It will be interesting how trust plays into all of this.

  38. 70
    Marcela Trujillo says:

    Hello Avinash,

    Currently, I am deepening in the area of Digital Marketing, improving myself in Customer Experience and Inbound Marketing.

    After reading some of your articles on AI and how it will impact some careers in the future, I started to look at this topic very carefully now. I do not know yet how I could introduce this knowledge into my field of study and use it professionally.

    I really want to understand more about AI and your articles are helping me get started on it.

  39. 71
    Darell Jaggers says:

    I believe your projections as to how long it will take for AI to really impact us is a lot sooner than your suggestions. I believe that as more information is collected and analyzed things will speed up.

    I was always lucky, I had directors that wanted me working to my potential. Many of my colleagues called me the spreadsheet and database man. I started working with computers in the late 60's.

    My CEO knew that I was creating formulae to determine the best processes for most efficient operation. He wanted to know why I was taking so much work home. I told him I could spend most of the day calculating and double checking things at work. I was able to reach the same conclusions in seconds at home.

    Two days later, I had the first PC in the plant. I was asked if I could take our old analog test stands into the digital world. I told them if the project engineer could give me the data for testing.

    The engineers only had other OEM parameters that they used for test purposes. They sent an mechanical engineer to work with me for half a day. Our first task was to evaluate and graph all the written data accumulated from the analog test stands. This allowed us to establish test parameters to try.

    I bought old Brunelli boxes to automatically capture the data from our newly computerized test stands. I used old Basic code for writing our algorithm. Eventually, we were able to predict what caused our failures. The more data we collected and analyzed the better our results and failures decreased sigbificantly.

    My next job was to create balanced production lines to improve the quality of the products being assembled. Supervisors hated me because of adding data collection to their jobs. The process being used was to pull all the parts for a product and place them on flats.

    The assemblers would go through and look at parts and say this goes in a subassembly and this goes in the final assembly. I went back to the engineering department and asked why our BOM's didn't use the computer system to pull parts in a configuration that did the sorting of parts for us.

    I was told to write up what parts I wanted picked when. I spent some time doing that task, but the tooling person agreed to help. I told the tooling supervisor that we would eventually need to ID the tool required to assemble each stage identified.

    The tools were scattered everywhere and many times the wrong tool was being used creating quality problems. We eventually created color coded carts that placed the tool required for each station on the line. Quality and time were both improved.

    Our production line set up required pre positioning of everything and was called out on our work order. The more we removed decision making from the production line the better. Each step improved both time and quality.

    We continued to tweak and improve our processes. The guys on the production line opposed me initially, until they determined many of the things they had complained about were being made easier.

    I told management that the people were really giving me some very good ideas and should be involved in preplanning of new products. I had a very good network around me and it made my job much easier.

    I always used scientific methods and probabilty in establishing processes. I've been retired for 17 years now, but still work on my three computers. I used to take college classes to try and stay up to date.

    I knew some NASA people and picked their brains to improve our processes and quality. Money wasn't usually to difficult for me to get, I had a good record of making good and profitable decisions.

    I have been suggesting that personnel from many areas of expertise should be evaluating items on the shelf today. There are several times that innovation in areas are stumbled upon by need. Then the industry adopts it for their personal use. I know for a fact, that many of today's problems have solutions available just not realized.

    As we gain knowledge and realize that there is so much more we don't know, than we know. We will see an exponential growth in many areas that will move us forward at a much faster rate than many realize. People used to tell me, I eliminated emotion from my decisions. I needed to think like a person not a computer.

    Logic has probably ruled much of my life and the more data I had made my decisions more logical. It seemed that I was making good decisions in less time than many others.

    AI through the use of sensors and continually improving algorithms with faster and faster computer speeds allows more processing in less times. All this being tied together collectively with the capability of more and more iterations will definitely move beyond human capability.

  40. 72

    Thanks Avinash,

    You've sufficiently freaked me out about the future of digital marketing. The potential of collective learning was eye opening and ML-driven attribution has me rethinking a lot of things.

    Thanks for sharing.

  41. 73

    Following up on your concept of Collective Learning, Avinash, I would like you perspective on why most retail companies are focused on trying to optimize their businesses leveraging just their own data rather then participate in cooperative digital databases. Just as clinics and hospitals which apply ML only to the data generate by their own patients are unlikely to achieve any significant insights, marketers who apply technology to their limited data sets are just making their camels go faster.

    In reviewing the Google Analytics solutions :
    https://www.google.com/analytics/resources/#?type_filters=marketing-analytics-and-attribution

    The focus from Google still seems to be helping clients optimize based on their own data. But the real power will come from the Collective Learning…

    In the 1980s catalog marketers realized that there was more to gain from Collective Learning than analyzing their own transaction and circ data. Co-op databases were transformative.

    The current marketing landscape seems to focus on 2nd party data being bought and sold, or exchanged on a 1:1 basis (eg the recently announced Google Walmart partnership).

    https://blogs.adobe.com/digitalmarketing/analytics/2nd-party-data-continuum-data-sharing-hot-right-now/

    Why isn't Google (via Adometry/Analytics) or another "clearing house" for data leading the development of Collective Learning for marketers by launching a digital database cooperative for 2nd party data which includes transaction, search, browsing, social, etc.

    Applying ML to 1st party cross channel data is making camels run faster. Applying it to 2nd party data (at scale) is driving a car.

    Or am I missing something?

    • 74

      Parker: You are not missing anything, Deep Neural Networks need huge sized classified datasets to learn and then optimize. Having more data would indeed be better.

      What is already possible is that platforms like Google will take your data (CRM, ERP, other data) and merge it with the behavioral data they already have and you can do transformative marketing and customer service. The other way also works, data from Google, Facebook, Verizon/Oath back into your datasets to optimize your offline experiences or your phone channels. Do this now if you are not already.

      What might be possible is to have Expedia, Walmart, American Express, Pfizer, and IBM merge their datasets to get a broader – collective – understanding of humans to do better marketing and customer service. I say this is possible because none of them truly are competitors and can see that there is a symbiotic relationship here. And, don't forget to do the above.

      What might not be possible is to have Nordstorm and JC Penny and Sears and Gap to create unified datasets. The benefit will be interesting, as you mention. Too much competition.

      -Avinash.

  42. 75

    Artificial intelligence is a modern direction and the future in it!

    Thanks for sharing this article!

  43. 76
    Sekhar Sarma says:

    Hi Avinash,

    I had been following your blog for some time, and have sufficiently got myself convinced & motivated (in that order) to work & build a career in Web Analytics.
    Coming from a non-Web non-coding background, my roadmap for this journey is, your advice from this post.

    https://www.kaushik.net/avinash/web-analytics-career-guide-job-strategy/

    Additionally, I also see your latest post on Artificial Intelligence, and the way it is changing the landscape of jobs & careers, as we know.

    https://www.kaushik.net/avinash/artificial-intelligence-opportunity-camel-to-cars-moment/

    For a newbie in Web Analytics (and a middle-aged person) who's trying to build a career and make a switch, your post definitely has 'freaked me out'.
    2 reasons:
    1. My traditional IT role, seems to be ripe for extinction, in < 5 years.
    2. My new-found love Web Analytics, into which I started investing my time & energy – may also have a shift in landscape.

    My request is, how does a person like me, can still go about building a career in Web Analytics/AI?
    What should be the typical opportunities/roles I should target in 1.5 to 2 years, so that I can be 'employable' for the next 10 – 15 years?
    What additional 'skills' should I acquire, especially – I am starting on this path, NOW…

    In summary my question is, the road-map as you've written in 2011, how changed would it be, if you were to write it today?

    Would that be a relevant enough question for you to respond, write a separate post?

    Thanks, in anticipation.
    Sekhar

  44. 78
    Todor Musev says:

    You might find this Ted talk useful: "The era of blind faith in big data must end"
    https://www.ted.com/talks/cathy_o_neil_the_era_of_blind_faith_in_big_data_must_end
    A lot of wisdom, good examples…and very, very important questions and concerns.

    Bes!

    • 79

      Todor: A really great talk, thank you for sharing it. There is no question GIGO continues to be the principle that still pays a pivotal role. GI will result in GO.

      On top of the examples Ms. O'Neil has shared, there are more that show how Unconscious Bias creeps into the classified datasets we use.

      I take Ms. O'Neil's call to action as something to incorporate into what we are all doing, I do not take it as a stop and run for the hills cry. :)

      And, the fantastic news is that all you and I are just trying to figure out how to get smarter data out of an analytics tool or not irritate a human with a irrelevant display ad.

      Avinash.

  45. 80

    Avinash, you are totally right :) This shit is real.

    A few days ago, when I open my Google analytics account, without me pressing even a single button, intelligent insights shows me 10 signals about the performance of my website along with the summary and some recommended insights. I didn't do anything still I can know at a glance how my website is performing.

    And intelligent insights will keep getting smarter and faster as the time passes. Definitely, people need to take it seriously and start focusing on improving their skills. AI creates both opportunities and risks, and we need to decide which way do we want to go.

    Thanks for another excellent post.

    With Regards,

    Bhola

  46. 81
    Vishwajeet Singh says:

    Interesting and yet very realistic take on the future of the AI.

    I totally agree with you.

  47. 82

    Hey Avinash,

    First Time over Here – was here to read about Smarter Career Choices but quickly jumped on this one because I was also from the lotto who seems more interested towards AI.

    Although we have nothing to do with it still I really want to read more about it & wanna to find it out that when exactly AI will start effecting human day-to-day Life

    What's your overview of it – will it going to cross the VR or Human Psychology with its effect?

    Cheers,

    • 83

      Rahul: There are many SiFi novels/stories where AI is running simulations that look and feel like real life… Except they are basically Virtual Reality. It is not hard to imagine with all the technology evolution that that is well within the realm of possibility.

      It is very hard to imagine what the implications on Human Psychology is. If you get my newsletter, The Marketing <> Analytics Intersect, you can checkout TMAI #100 in which I'd outlined three slightly overlapping possibilities for humanity – looking 150 to 200 years out. That might be of value.

      Avinash.

Add your Perspective

*