Quote of the day 21st August 2024

 “For a Remainer like myself, it’s painful to admit — but when it comes to AI regulations, it’s far better we chose to 'take back control'”.

"Sadly, in the face of overwhelming evidence that their top-down approach" (to digital and AI regulation) has failed, EU leaders haven’t tried a different tack, such as market-oriented reforms to reduce barriers for digital businesses. Instead in 2022, Brussels introduced the draconian Digital Market Act, which has clobbered tech companies with fines worth tens of billions of euros.

But the pièce de résistance is the EU’s AI Act, hastily drafted and passed this year in response to the public debate sparked by the launch of ChatGPT. Breathlessly heralded by European leaders as “the world’s first AI legislation”, it’s genuinely hard to overstate how dreadful for innovation and growth these rules will be. For starters, the legislation thickheadedly clamps down on AI models in general (which can be used in all kinds of ways), rather than more sensibly focusing on potentially harmful specific applications of AI, such as deep-fake pornography videos.

Bizarrely, the EU has set an arbitrary upper limit on the size and power of these AI models — and in such a rapidly progressing area, that ceiling has already been exceeded by the latest AI systems, meaning these can’t be used in Europe. We’re only at the beginning of the AI age, so setting blanket limits like this is as barmy as a 1960s legislator setting an upper limit on the processing speed allowed in computer chips, with no way of knowing how the technology would develop in the following decades.

On top of this, the EU is introducing curbs on AI in fields like healthcare and medicine, as well as imposing onerous requirements on developers to provide users with detailed summaries of the content used to train AI models. EU officials admit the red tape burden will be significant for small firms, easily adding up to six-figure amounts for startups. No wonder the EU bureaucrat chiefly responsible for the AI Act has confessed to Bloomberg: “The regulatory bar maybe has been set too high… There may be companies in Europe that could just say there isn’t enough legal certainty in the AI Act to proceed.”

"The rushed rules are so patchy and the AI field so fast moving, it’s thought a further 60 pieces of secondary legislation will be needed to plug the gaps. This will have a chilling effect on AI innovation." 

"Even though Britain is outside the EU, this matters to us. The EU is explicitly trying to impose its AI regulations on companies worldwide. If a UK AI company has a single user in the EU, it has to comply wholesale with the legislation — and the significant compliance costs involved — or face severe financial penalties. This will hamper the growth potential of British businesses — not good news for our economy."

"And because the EU’s halfwitted AI rules are supposedly about “safety”, this puts pressure on British politicians to follow suit or risk being blamed whenever AI causes problems. Hopefully Sir Keir Starmer can keep his head on this issue, but it won’t be easy to resist demands to simply use the barmy EU laws as the basis for UK regulations.

The economic cost of following the EU down the wrong path on AI could be monumental. Willem Sundblad, the Swedish founder of AI startup Oden Technologies, told me: “The only thing the EU Act will ensure is that no meaningful AI innovation will come from the EU compared to the US. European AI businesses will be hiring lawyers and lobbyists, while the Americans recruit software engineers.”

"What a shame it would be if Britain undermined its nascent AI sector with dunderheaded EU-style rules."

Extracts from a Times article by Rohan Silva, "We can’t let EU’s anti-AI rules drag us down too."

You can read the article in full by clicking on the link below:

We can’t let EU’s anti-AI rules drag us down too (thetimes.com)