Lost your job at OpenAI? Become a tech regulator

YOU rarely see any budding entrepreneur on Shark Tank or The Apprentice say their ambition is to be a government regulator.

Yet that is pretty much what the European Union’s (EU) freshly unveiled AI Office is counting on, as it hires 140 staff –including technology specialists – to enforce the 27-country bloc’s landmark artificial intelligence (AI) regulations, the world’s most extensive. The regulator’s to-do list is a long one: Impose guardrails on a relatively nascent technology, test and evaluate general-purpose AI models, issue fines as needed and also promote “trustworthy” AI to unlock the productivity benefits companies are hungry for.

But with a reported €46.5 million (S$68.2 million) starting budget in a field where Elon Musk can raise US$6 billion in one funding round, and even the likes of French President Emmanuel Macron critical of Europe’s tendency to over-regulate in technology, what hopes are there of actually hiring so-called “Oppenheimers”?

There are reasons to be optimistic. The need to regulate AI, and regulate it well, is spreading in the public consciousness. Concrete issues are taking the place of exaggerated sci-fi depictions of what the technology could one day do. Generative AI’s real-world harms include its use in scams and bank frauds, sowing election misinformation and generating sexually explicit and abusive fake images of real people, from young girls to pop star Taylor Swift. Artists and creators like Scarlett Johansson and Billie Eilish are publicly complaining of their likeness being ripped off by bots before they can even object. Polls show a majority of Americans favour AI regulation.

Even inside the tech world, despite the dizzying AI hype driving the likes of Nvidia and Alphabet to record market capitalisations, there are increasingly public signs of disillusionment. What began as a chorus of industry voices worrying about the existential risks to humanity from AI in the long run has become a more urgent and immediate concern. Last year, AI music expert Ed Newton-Rex quit high-profile startup Stability AI over its view that using artists’ work without consent could be considered “fair use”. The former executive threw his support behind “serious regulation around the world” and created a non-profit, Fairly Trained, focusing on AI transparency.

I asked Newton-Rex about the positive impact industry insiders could have on helping to directly regulate AI. “I worry things will get missed if we don’t have experts involved in regulation,” he told me. Newton-Rex cited the risk of unlicensed data being used to train AI models or of synthetic (or AI-generated) data used from other sources that infringe on copyright. He also raised concerns around the use of opt-outs, which are designed to protect rights holders’ work from unauthorised use. The idea of making less money but doing more good may be spreading; in theory, Europe should be exactly the kind of place where smart well-off tech gurus turn their thoughts to the public interest after getting rich.

SEE ALSO

If AI can do your job, maybe it can also replace your CEO

There are caveats, of course. History is littered with examples of regretful innovators in tech, from the WhatsApp co-founder who joined the #DeleteFacebook movement after falling out with Mark Zuckerberg to Jack Dorsey’s ambivalence about Twitter’s legacy. That does not necessarily make them a good fit for the staid world of government institutions – the head of the AI Office has worked for the EU since 1997. It also potentially raises the risks of unintended consequences like regulatory capture or diminishing returns. One 2014 report found that a financial regulator hiring quant experts to better level the playing field against banks had later run into the unexpected problem of too many quants locked in vehement disagreement.

Still, given the deep pockets and big egos of the Promethean billionaires chasing AI riches, regulators have little choice but to try to seize the opportunity of finding the kind of person who understands this tech yet is also willing to say no to a five- or six-fold bump in salary. Whether that will be enough to turn Europe into an economy that does not just regulate but can actually create world-beating AI champions – the other kind of Oppenheimer that Macron is rooting for – is a whole other story. 

The writer is a Bloomberg Opinion columnist writing about the future of money and the future of Europe