Is Artificial Intelligence COVID19’s Latest Victim?

Before the pandemic and economic upheaval caused by COVID19, the top issue dominating the tech news was Artificial Intelligence. The major subtopic that dominated the zeitgeist was: will AI automate your job?

In short, before the pandemic, there was a low likelihood that net job loss from AI would be that high. While a lot of jobs probably were going to be automated in the coming decade, AI would also create a lot of better, well paying, jobs. The problem of course is that the people displaced by that automation most likely won’t have the skill sets for the new jobs created, causing tremendous social issues. In early 2020, before the pandemic, AI was not mature enough for mass automation but rather positioned to augment knowledge workers because AI is:

• mostly scoped to machine learning techniques focused on prediction. While prediction can do some amazing things (power a self driving car, preform facial recognition) it is not intelligent enough to make edge case decisions, limiting its application and mass automation capability
• only as good as its underlying training data and as we learned from the racist and sexist facial recognition debate, garbage in, garbage out still applies.

It would take years, if not decades, for AI to be ready to begin to automate a significant amount of jobs. (AI augmenting and transforming lots of jobs is still a very real thing.) Broadly speaking, AI needs more maturity and some more breakthroughs (as well as a solution to its data problem) before it will be in a position to automate a lot of jobs.

AI Winter?

After the world recovers from the COVID19 pandemic and recession, AI will be in an even weaker position to automate jobs then before the pandemic. There are three reasons: funding, FOMO, and politics.

Funding

AI is a technology that is fueled by research. Academics way smarter than me at universities and corporate research labs have amazing breakthroughs on AI and write papers and books on them. Then we in the tech industry apply those techniques commercially and push the limits of the breakthrough. As these breakthroughs happen inconsistently at the pace of the academics, the progress of AI has been lumpy. Compared to the consistency of Moore’s Law, AI has had several “AI Winters” of no breakthroughs. As AI research is funded by universities, government grants, and R&D budgets of major corporations, during AI winters, funding dries up. As a result, less people go into the field and the time it takes for the next breakthrough is even longer than under normal circumstances.

Funding for AI is about to dry up in a major way. As universities, governments, and even large tech companies flush with cash struggle to cope with the economic shutdowns, recessions, and massive job loss, funding for AI will not be a top priority over the next few years.

VC FOMO

AI was a super hot sector prior to COVID19, there are even stories of an underwear company positioning themselves as an AI company. Venture Capitalists tend to invest as part of a herd mentality for the fear of missing out (FOMO). VCs rush into a sector because its hot. Until its not. Remember Virtual Reality? Crypto? As fast as a sector becomes hot, it can also become cold. When it becomes cold, VCs ignore it and move on to the next big thing.

Even as billions of dollars have been invested into AI over the past few years, AI is not playing a key role in the pandemic. The inconsistency of the COVID19 testing methodology and data worldwide (garbage in, garbage out) and the lack of scientific knowledge about the coronavirus itself, will make the doctors and medical professionals the heroes of the crisis, not AI. The biggest supporting role in this crisis is Zoom, and other work from home tools. In our darkest hour, a medical degree, DNA sequencing technology, and video conferencing software outweigh a neural net.

These are the new sexy sectors for VCs to get FOMO over: health care and future of work. While some underlying AI can play a role in those two sectors, the mindshare will start to focus away from AI and towards solutions in other industries, thus depriving young AI startups of the capital they need to push the boundaries of AI further. While AI will power some of the digital health and future of work solutions, however, it will be part of the supporting cast and not push AI past its boundaries.

Politics

As the governments around the world bailout industry after industry and provide tremendous financial support to banks, to combat unemployment levels never seen before, there will be no appetite by governments to allow jobs to be automated by companies that took government assistance. Remember the backlash after the too big to fail bank bailouts in 2008? The current COVID19 crisis is bigger and politicians have seen what the outrage would look like. The politics around the pandemic will set back AI, potentially for a generation.

While AI may be headed for a COVID19 induced AI winter, the momentum behind AI was the strongest it has ever been doing into 2020, so I suspect it will emerge from its winter faster than in the past.