In anticipation of the ethical challenges that would present themselves as a result of advancements in medicine, biology, and technology, the field of bioethics was born, an interdisciplinary field of study whose aim is to provide frameworks for making decisions in a world defined by increasingly advanced, complex technology. For example, how does society respond to the bioethical dilemma presented by cloning? History and humans have largely (so far) demonstrated there is a line they don’t want to see crossed.
Fast forward to this century, and of late, there’s been ample discussion of “Artificial Intelligence” (AI) lately. Over the past few months, major global technology companies from Amazon, Facebook, Microsoft, IBM, and even Apple have “open-sourced” elements of their technology stack to make them available for developers to freely contribute to and build on top of. And, most recently, we’ve seen the launch of OpenAI, a billion-dollar-plus nonprofit initiative with the aim of advancing AI to make its discoveries open as a check against the larger, aforementioned companies having a controlling monopoly on these technologies.
The topic of AI itself is massive, a life’s work only a few people with the proper, interdisciplinary training can even hope to master. I am not in that league, but I do observe how the term has been thrown social media. People now see and hear the term “AI” thrown into tweets, slide decks, email blurbs, press releases — the language of AI has taken on a life of its own. To make sure I understood the issues, I found it useful to read through resources like this, and what I was struck by was how multiple advances in seemingly unrelated technologies (such as drones and neural networks) could create a moment in time where AI, as a field of study, could cross the threshold from siloed experiments into integrated systems in the real world.
And, if we (1) look back at the field of bioethics and issues such as cloning, we start to see that humans have arisen to place parameters on what advancements in integrated technologies could afford; and, again, if we consider (2) the growing strength of public Internet companies such as Apple, Amazon, Facebook, Google, and more, it becomes easier to envision a world where these companies grow even bigger and stronger; where they extend their product lines deeper into mobile phones, drones, satellites, and other devices connected to the web; where they continue to accumulate rich, dynamic, never-seen-before and unreplicable data sets on our behaviors, intents, relationships; and corporations which, thanks to the glorious network effects of the Internet, be awash in cash profits for decades, to the point where they can afford to open-source some of their most core technological insights.
In a world where resources for AI and machine learning are open-sourced by the larger incumbents, the impending democratization of those technologies will likely lead to a situation where the most proprietary value lies less within AI, but more within the corresponding datasets the machines need to process information from in the first place. Building new products and services with open-source technology may help foster and accelerate innovation, but who will own or have access to those datasets? It’s a question I’ve been thinking about as we will likely see a whole new wave of entrepreneurs who try to harness these free services in the same way they did with platforms like iOS, Android, and the advent of cloud computing.
Seen through this lens, the moves by the big tech companies and the creation of OpenAI all make perfect sense. It may seem like a lot of money, but this is just the beginning. Machines are increasingly mobile, increasingly able to handle more tasks, increasingly able to more efficiently harness the power and flexibility of the chips and operating systems they sit on top of, and are increasingly lowering in BOM and operational costs. One of the world’s premier VC firms, which publicly discuss its investment thesis as one where “engaged networks of people can disrupt large markets,” has invested in a company in 2015 where the thesis is extended to a network of machines can not only take the place of people — the machine network can operate 24/7, learn with each new piece of data it collects, and become its own platform for other developers to build on top of.
I am less of a long-term thinker these days. I think about things that could happen 3-5 years out. Or, I try to. But based on a few investments I’ve made in the space, the increase in pitch meetings where these integrated technologies are brought to market, and the timing around all the larger companies open-sourcing some of their secrets, it strikes me that a world of intelligent machines — Computer Sapiens — may arrive much sooner than we collectively recognize. Already, machine learning products are improving interpersonal communications, sales and marketing automation, and many more, all along the way driving margins by lowering the need for human input and maintenance. Who owns and who has access to not only these technologies, but also the datasets that will drive them, will be a fascinating fault line to observe as the plate tectonics of AI shuffle over the next decade.