It’s time for a new paradigm in AI research

The current trajectory of AI falls short for business and humanity

By Paul Lee, M.D., founder and CEO of Mind AI

Depending on who you ask, artificial intelligence is either everywhere or nowhere in 2018.

If you’re using the popular definition, you’d say most people interact with AI multiple times a day through voice assistants, facial recognition, targeted advertisements, search engine algorithms, and many other applications. But if you’re using the Defense Advanced Research Projects Agency’s (DARPA) definition, none of these applications actually represent true intelligence.

They’re all just sophisticated spreadsheets.

None of the AI systems on the market today actually ‘think’ in a way that resembles human thinking. Their capabilities rely on deep learning, which involves feeding massive datasets into supercomputers to train a program to recognize patterns. The domain expertise of these machines is very narrow, and their expertise is not transferable — they all require retraining to gain expertise on any new subjects.

Artificial general intelligence, which DARPA calls the third wave of AI, will be reached when machines can reason logically and adapt their knowledge to different contexts. However, in pursuit of this goal, most researchers in this field focus on improving deep learning — because that is all they know. They believe that whoever can crunch the most data the fastest will unlock the keys to real intelligence. This approach is driving up the costs of using AI because it requires massive computational power, which is expensive to own and operate or to rent by the hour.

Based on the expectation that all further development in AI will only demand more computational power, a massive amount of venture capital investment is flowing into hardware startups. In 2017, VCs pumped more than $1.5 billion into chip startups, according to research from CB Insights.

The traditional approach to AI relies on massive computational power.

Thanks to the significant barriers to entry into this field, a few giants have begun to concentrate the talent and financial resources needed to continue pursuing this approach to AI, leaving others in the dust. The already big names in tech are turning out to be the big names in AI — Google, Amazon, and Facebook in the U.S., and Baidu and Alibaba in China.

Today’s AI misses the mark for business and humanity

The corporate world has been told that AI will be the magic bullet that solves everything, but most businesses are struggling to find meaningful ways to put it to use. In a previous startup, I had access to IBM Watson, and was excited to harness the tool to help human doctors and veterinarians make more informed diagnoses. But when my team explored its actual applications, we came up short.

Humanity as a whole stands to lose out with the traditional approach to AI development. If development continues to be concentrated in only a few hands, taking a closed-source, capitalist approach, access to breakthroughs in automation and productivity will be limited to those who can pay for it. And when only the rich and powerful have access to life-changing tools and information, wealth and health inequality gaps widen.

“Imagine if skills could just be downloaded — what’s going to happen when we have this kind of AI but only the rich can afford to become cyborgs, what’s that going to do to society?” — Shimon Whiteson, an associate professor at the Informatics Institute at the University of Amsterdam, in a Business Insider article.

AI tends to inherit biases from its developers. A few studies have shown the harmful impact of these biases, for example, facial recognition systems failing to distinguish between dark-skinned individuals with the same accuracy as light-skinned. If the development community remains undiversified, we can expect the systems they create to continue to exhibit the biases of the world’s elite, educated, and privileged, creating bad user experiences or even dangerous situations for minorities who interact with those systems.

While the 3rd wave of AI could unlock many positive changes for humanity, it could also bring about even more powerful weapons of mass destruction. Last year, Russian President Vladimir Putin made an ominous pronouncement about AI and autonomous weapons, saying that “whoever becomes the leader in this sphere will become the ruler of the world.” The power of dictators will go unchecked unless the public has access to the same kind of technology they do.

When we think of AI today, we tend to focus on the wrong things. We jump from one extreme to another, either giving into fears derived from pop culture doomsday scenarios, or doubting humans’ ability to ever create artificial general intelligence. But what we should pay the most attention to is how we go about researching and developing this technology, because the next wave of AI will change everything. We need to alter course from the current trajectory of closed-system, competitive development to make sure that the life-changing benefits of AI are shared equally across the planet.

We need a new paradigm in AI research to safely unlock the power of AI for all. That’s why we’re building Mind AI.

To stay up-to-date on our progress, sign up for our email list or follow us here on Medium.