The companies that broke AI may be the ones to fix it
Illustration: Aïda Amer/Axios
Just 9 rich tech companies have come to dominate the development of AI. In the process they have made monumental errors, leading to privacy meltdowns and biased algorithms.
In a new book, New York University professor Amy Webb stakes out the controversial position that these companies — though they stirred up the trouble — are also the best hope to fix it.
What's going on: In "The Big Nine," Webb profiles 6 companies in the U.S. and 3 in China.
- The Americans are a familiar bunch: Google, Microsoft, Amazon, Facebook, IBM and Apple. Webb calls them the G-MAFIA.
- China's big 3 are Baidu, Alibaba and Tencent.
- Together, the 9 are valued at about approximately $4.9 trillion, and they employ most of the world's still-slim ranks of top AI talent, numbering just in the thousands.
In the U.S., the G-MAFIA has driven the development of AI, putting out razzle-dazzle products every year in an attempt to appease investors and shareholders.
- This has meant that the companies have shirked important long-term planning for AI catastrophes, Webb says.
- Now they are under fire, taking hits in the eyes of regulators and the public for their rapid-fire missteps.
So why give them any more power than they already have?
In an argument against throwing the baby out, Webb tells Axios that the companies, with their top talent and vast institutional knowledge, are the best vehicle for fixing this mess — as long as they have a firm hand to guide them.
- They can't be broken up like Bell in the 1980s, she says: "It's totally impractical."
- The normal course of regulation, too, is a no-go, she argues — too slow to keep up with the rapid pace of change.
- Instead, what's needed is an international oversight body with enforcement power, Webb says — something like the International Atomic Energy Agency, which polices countries' nuclear programs, but for AI.
If they stay the course, the G-MAFIA are setting up themselves — and us — for catastrophic failure.
- At the root of the problem is that their labs, which are largely homogenous — are made up mostly of white, highly educated men.
- AI systems these groups produce have been shown to treat some groups, like women and people with darker skin, differently than others.
- A new digital divide is widening between people who look and think like these AI makers, and the forgotten others, Webb tells Axios.
As American companies are left to their devices, China is sinking huge money into AI development, as we've reported. Considering the collision course between the countries, Webb plays out three scenarios:
- In the optimistic vision, the G-MAFIA form a coalition with shared objectives, drawing in more and more participants until China is pressured into adopting its norms.
- The neutral scenario in which things continue as they are, humanity adjusts to living with "millions of paper cuts," the result of thoughtless, uncoordinated AI development. The companies consolidate into a few mega-corps, and the lives of haves and have-nots diverge drastically.
- The disastrous pessimistic outcome sees China's authoritarian regime rise above the rest, far surpassing a U.S. that never moves past the issues of the 2010s. Beijing, in possession of super-intelligent AI and with no further use for non-allies, slowly exterminates the hapless West.
Drawing inspiration from the Bretton Woods conference that birthed a new post-war financial order, Webb suggests creating a new international body to keep the companies in check, and to preserve human rights as they develop more powerful AI.
- China's authoritarianism will only be balanced by economic forces, Webb writes. By funding basic AI research in the West and pushing the G-MAFIA forward, the world can avoid fantastical disaster.