How to think about the artificial intelligence wave

Doug Clinton, of Loup Ventures, laid out a framework for “how to create most value in the next technology wave” on Techcrunch. The Loup ventures framework for value (paraphrased) is -

  • Operation System: If you can create an operating system — defined as combining core pieces of the new technology into a single platform that makes them available to third parties, create an OS. Examples: Microsoft for PC, Google for the web, Amazon for e-commerce, Facebook for social, iOS and Android for mobile, etc.
  • Killer user case of a component: If you can’t create an OS, find a killer use case of a component of the new technology platform. E.g. Uber (location) and Snap (camera) for mobile, Dropbox for file sharing on web
  • Focus on one untouched industry: If you can’t optimize a component, disrupt an industry untouched by the technology. E.g. Uber again (taxis), AirBnB (unused homes), SoFi (loans)

I thought it was a useful framework (thanks Doug) and one we could use to examine how to think about the artificial intelligence wave we are in.

The AI Operating System.
While it is unclear if a go-to operating system has appeared yet, it looks like the AI wave will have an operating system that looks more like Android than iOS. Google were quick to open source TensorFlow — an open source library for machine learning — and Tensor2Tensor — for deep learning. So, Amazon, Microsoft and Facebook followed.

It is unclear which of these will emerge as the de-facto platform but it is clear that all of these will play big roles as the AI wave unfolds.

Why did we end up with operation systems that are free? I think the biggest reason we ended up free is because Google open sourced TensorFlow and, in some ways, created the rules for the market. It also makes business sense for the AI OS to be free. The differentiator in creating awesome AI based applications is data. And, Google has plenty of that. The company has also built a culture around making sense of vast amounts of data. While that culture was not suited to the social wave, it is perfect for artificial intelligence.

So, Amazon, Facebook and Microsoft were forced to follow.

Apple also followed with a new deep learning framework for iOS. But, it has had its struggles as the culture of the company is suited to developing awesome, high margin devices. Culture is strategy in the long term, after all.

Of course, there are lots of start-ups that are competing with the big five. Paperspace, for example, is focused on GPU powered virtual machines for data scientists. Clarifai focuses on visual search. But, the giants are going to be hard to beat given their massive data advantage and their commitment to investing heavily. Google, for example, just created a venture fund focused on AI and Microsoft announced a lab to take on Google’s Deepmind.

The game is on.

A China note.
There are 2 internets for the purposes of any discussion, the Chinese internet and the “rest-of-the-world” internet. Most of my notes tend to be focused on the “rest of the world” internet. This isn’t different. Just as the big five of Amazon, Apple, Google, Facebook and Microsoft dominate in the rest of the world, Alibaba, Baidu and Tencent are likely to dominate in China.

The Home OS / The Voice OS
The interesting thing about the operating system definition in the Loup ventures framework is that they don’t view it as a single OS per wave. They define it as “core pieces of a new technology into a single platform that makes them available to third parties.”

So, maybe we need to widen how we think about an OS for the AI wave. For instance, we are in the midst of an interesting battle to become the de-facto home OS / voice OS (depending on your point of view). And, I saw this promo on the Google homepage yesterday to “get Google Home.”

This is a HUGE deal. Google rarely promotes anything on its homepage. And, it is likely a reaction to Amazon’s self created Prime day.

Amazon looks like it is winning the battle for the home.

Maybe that battle will end up being separate from the battle to be the voice digital assistant of choice…

Killer use cases and industries that could be disrupted.
Once we get past the difficulty of challenging the giant technology companies to define the operating system, we are now left with the second and third sources of value — killer use cases and industry disruption. This is where things get really interesting as there are literally hundreds of start-ups coming at this from various angles. They also have an advantage here as it is hard for the bigger tech companies to specialize in every niche. We’re still in the early days here and there aren’t obvious example of killer applications that have created enormous economic value — yet.

But, it raises an interesting question — What is a framework that would help us make sense of what artificial intelligence is capable of?

I don’t yet have a framework but we have a list of broad themes thanks to an excellent video by Frank Chen at a16z on “The Promise of AI.” Frank approaches this with the question — what will AI make cheap?

He began by comparing the AI revolution with that of relational databases. Relational databases made it easy to store, sort or count information and these, thus, became fundamental in every piece of software.

(Example of relational database thanks to Envato Code Tutorials)

So, what will AI make cheap? AI will…

  • Enable machines to move place to place: Consider use cases including robot powered delivery to drone fire fighters to drones delivering critical stuff — e.g. blood in Africa (this is happening already).
  • Enable machines to see and understand the world: We are already seeing this via visual search on Pinterest. But, there are some powerful applications when it comes to sorting things, recognizing the variants and doing something with the data. A great example is Blue River Technologies, an agricultural technology company, that can spray the right amount of water per lettuce leaf — based on its needs.
  • Create content: We’ve seen applications like machine generated press coverage, cookbooks, music, movie trailers, and software.
  • Use large amounts of data to optimize complex systems or predict the future: Think recommender systems, better customer service systems as companies can connect experience and pick up from where you left of in interacting with them.
  • Understand people and be understood: Applications like instant language translation, context based suggestions for counsellors, doctors, attorneys and even email writers (Smart replies by Google) and automated summaries.

(There is, of course, more to this than Frank’s list. But, his list is a great start and beautifully points to how we’re seeing machines do things we do, better.)

Here’s another way to visualize what AI enables machines to do.

As you might imagine, this will affect the livelihoods of a lot of people as beautifully illustrates.

While we undoubtedly need to do a better job preparing for this future, an interesting question that follows is — what kind of work will we do? Or, put differently, what type of intelligence will we reward?

What intelligence do we reward?
For centuries, we rewarded prowess of the human physique. Our notion of “work” involved hard physical labor. Then, we began rewarding the sort of intelligence that we came to represent as “IQ.” This was left brained and logical and focused on memory and processing power. Over the past couple of decades, we’ve begun rewarding the right brain — creativity is seen as a powerful asset in today’s information based economy.

What happens next? Aeon has an interesting essay on how the future could reward emotional jobs that are currently grossly underpaid — think teachers and geriatric care workers. It is a compelling read (link below) and here are the last three paragraphs —

“There’s an enormous opportunity before us, as robots and algorithms push humans out of cognitive work. As a society, we could choose to put more resources into providing better staffing, higher pay and more time off for care workers who perform the most emotionally demanding work for the smallest wages. At the same time, we could transform other parts of the economy, helping police officers, post-office workers and the rest of us learn to really engage with the people in front of us.

This isn’t something our economic system, which judges the quality of jobs by their contribution to GDP, is set up to do. In fact, some economists worry that we haven’t done enough to improve the ‘productivity’ of service jobs such as caring for the elderly the way that we have in sectors such as car manufacturing. Emotional work will probably never be a good way to make money more efficiently. The real question is whether our society is willing to direct more resources toward it regardless.

Technology-driven efficiency has achieved wonderful things. It has brought people in developed countries an astonishingly rich standard of living, and freed most of us from the work of growing the food we eat or making the products we use. But applying the metric of efficiency to the expanding field of emotional labour misses a key promise offered by technological progress — that, with routine physical and cognitive work out of the way, the jobs of the future could be opportunities for people to genuinely care for each other.”

Being humane is defined as having or showing compassion or benevolence. Previous technology waves have never managed to align wealth and success with being humane. In fact, you could argue that being humane actually worked against you — as a warrior, a property owner, a businessman in the nineteenth and the twentieth century, or a stock market investor.

By having machines do the many things we considered uniquely human, it is possible that the AI wave will force us to reward being “humane.”

It is an outcome I’m rooting for.

Links for additional reading

  • How to create most values in the next technology wave — on Techcrunch
  • Deep learning open sourced — Google, Amazon, Microsoft, Facebook
  • Why Apple is struggling with being an AI powerhouse — on Washington Post
  • Paperspace launches GPU powered virtual machines for data scientists — on Techcrunch
  • New AI initiatives from the giants — Gradient ventures on Google Blog — and the new Microsoft AI lab on Bloomberg
  • China — Tencent has a massive data advantage thanks to WeChat on Quartz, Baidu’s latest AI investment on Techcrunch
  • The promise of AI by Frank Chen on a16z (HIGHLY recommended)
  • Future could involve rewarding emotional jobs on Aeon

This is an edition of a bi-weekly technology newsletter called Notes by Ada. If you like this and would like free weekly notes via email, please just subscribe here.