AI analysis: U.S. innovates, Europe regulates.

Artificial Intelligence will hold the attention of legislators and other interested parties for some of the final 15 days of the regular session of the legislature. CT Mirror’s Mark Pazniokas reported the challenges facing state Senator James Maroney as he tries to find a consensus on the way forward. Governor Ned Lamont may understand Connecticut could be diminished by the state becoming known more for its regulations than its opportunities.

The challenges of regulating expanding AI are being addressed by optimists and dystopians around the world. Jim Armitage, business editor of The Sunday Times, explained the dramatic differences he sees in approaches between the United States and Europe.

He wrote:

The US is leading the way on capitalising, and creating, the technology so far. Primarily, that’s because it has the most money and the biggest concentration of brains to do so. But also, it’s down to the lighter touch its policymakers take against the private sector as it explores and invests.

Stanford University’s Artificial Intelligence Index Report last week showed that EU regulators enacted no fewer than 32 AI rules last year. In 2021 it was 46, an extraordinary achievement of bureaucratic productivity amid the distractions of the pandemic.

Yet in the US, the global cradle of AI breakthroughs, they pushed through a paltry 25 and that was Washington’s highest ever. Meanwhile, and perhaps as a result, since 2017, US investment in AI has outstripped the UK and EU’s combined efforts in all but one of 25 sectors, from healthcare and financial technology to customer services. The exception? That red tape lover’s haven: data protection monitoring.

While we may be able to boast of being the safest regions in the western world, Britain and Europe also risk being the most technologically stifled. Where America innovates, we regulate.

The latest regulatory assault on the industry comes from Britain’s Competition & Markets Authority (CMA). Far from hiding its head under the duvet following its wrongheaded banning, then approval of, Microsoft’s merger with games maker Activision Blizzard, it is now blustering about how anti-competitive it is that big US tech companies are — horror of horrors — working with each other on AI.

Well, there’s a reason for that. The American tech giants are collaborating because building AI models is so vastly expensive — and financially risky — that it wouldn’t otherwise happen. Some context:

Microsoft-backed OpenAI’s GPT-4 upgrade used $78 million of computing power (or “compute” in the jargon) to train itself, while Google’s Gemini Ultra used $191 million.

Between 2017 and 2023, America — largely via its tech giants — spent some $18 billion on AI infrastructure compared with the EU and UK’s $100 million, the AI Index Report says.

So, when the CMA darkly proclaims that it has uncovered 90 AI partnerships and strategic co-investments involving Google, Apple, Microsoft, Meta, Amazon and Nvidia, I say: good. Bring it on.

Monopoly regulators across the western world are appalled that back in the day they dozily allowed Mark Zuckerberg to buy Instagram and WhatsApp. But those deals were about swallowing up smaller rivals’ existing technologies. This is now about pooling the vast resources needed to build the infrastructure of a completely new technology. One that could speed cures for cancer and the common cold, which is more than can be said for the worrywarts at the CMA.

Rather than try to stifle this new infrastructure, or vainly attempt to block its access to British and European markets, we should embrace it and invest in figuring out how to use it to create products and services the world will love, and pay us handsomely for.

Published April 23, 2024.