Deep Learning and Machine Intelligence Will Eat The World
Adam Rifkin stashed this in Artificial Intelligence
Shivon Zilis writes:
(originally published by O'Reilly here, this year in collaboration with my amazing partner James Cham! If you're interested in enterprise implications of this chart please refer to Harvard Business Review's The Competitive Landscape for Machine Intelligence)
Almost a year ago, we published our now-annual landscape of machine intelligence companies, and goodness have we seen a lot of activity since then. This year’s landscape has a third more companies than our first one did two years ago, and it feels even more futile to try to be comprehensive, since this just scratches the surface of all of the activity out there.
What's the biggest change in the last year? We are getting inbound inquiries from a different mix of people. For v1.0, we heard almost exclusively from founders and academics. Then came a healthy mix of investors, both private and public. Now overwhelmingly we have heard from existing companies trying to figure out how to transform their businesses using machine intelligence.
Machine intelligence business models are going to be different from licensed and subscription software, but we don't know how. Unlike traditional software, we still lack frameworks for management to decide where to deploy machine intelligence. Economists like Ajay Agrawal, Joshua Gans, and Avi Goldfarb have taken the first steps toward helping managers understand the economics of machine intelligence and predict where it will be most effective. But there is still a lot of work to be done.
On a lighter note, we’ve also heard whispers of more artisanal versions of machine intelligence. Folks are doing things like using computer vision algorithms to help them choose the best cocoa beans for high-grade chocolate, write poetry, cook steaks, and generate musicals.
We see all this activity only continuing to accelerate. The world will give us more open sourced and commercially available machine intelligence building blocks, there will be more data, there will be more people interested in learning these methods, and there will always be problems worth solving. We still need ways of explaining the difference between machine intelligence and traditional software, and we’re working on that. The value of code is different from data, but what about the value of the model that code improves based on that data?
Once we understand machine intelligence deeply, we might look back on the era of traditional software and think it was just a prologue to what’s happening now. We look forward to seeing what the next year brings.
This taxonomy started in 2014:
Shivon Zilis, an investor at BloombergBETA in San Francisco, put together the graphic below to show what she calls the Machine Intelligence Landscape. The fund specifically focuses on "companies that change the world of work," so these sorts of automation are a large area of concern. Zilis explains, "I created this landscape to start to put startups into context. I’m a thesis-oriented investor and it’s much easier to identify crowded areas and see white space once the landscape has some sort of taxonomy."
What is striking in this landscape is how filled-in it is. At the top are core technologies that power the applications below. Big American companies like Google, IBM, Microsoft, Facebook and China's Baidu are well-represented in the core technologies themselves. These companies, particularly Google, are also the prime bidders for core startups as well. Many of the companies that describe themselves as engaging in artificial intelligence, deep learning or machine learning have some claim to general algorithms that work across multiple types of applications. Others specialize in the areas of natural language processing, prediction, image recognition and speech recognition.
For the companies that are rethinking enterprise processes like sales, marketing, security or recruitment, or for others that are remaking industry verticals, the choices of technologies to license are dizzying. As Pete Warden, creator of the open source Data Science Toolkit, wrote in a recent post on deep learning, "I don’t see any reason why the tools we use to develop… and train networks, should be used to execute them in production." Entering 2015 we see all of this research finding its way into actual applications that relatively ordinary humans will use. "I also think we’ll end up with small numbers of research-oriented folks who develop models," Warden continues, "and a wider group of developers who apply them with less understanding of what’s going on inside the black box."
These companies will need more people who can create, iterate and debug deep learning and other kinds of machine learning models. They will also need an even larger cohort of developers and designers who can create usable experiences on screens that make all of this intelligence actionable. Big companies are poised to be the big winners here. Obviously they have the resources to attract or acquihire this talent. Even more crucial, big companies have big data and ongoing relationships with large numbers of customers. In machine learning, it is most often the quality and quantity of data available that is the limiting factor, not the cleverness of the algorithms.
And what most concerns the big tech companies from Apple to Google to Microsoft and IBM? Yep, mobile, and as Zilis points out, "Winning mobile will require lots of machine intelligence." Siri and Google Now are responses to the need for highly contextual voice interaction in mobile. Visual search like Amazon's FireFly involves location-based pattern recognition to create a pleasing experience. The reason for the current great enthusiasm for deep learning is that these kinds of problems can be solved now in minutes or days instead of years.