Trust us, says EU, our AI Act will make AI trustworthy by banning the nasty ones The European Union (EU) on Saturday reached provisional agreement on the AI Act – a broad legal framework limiting how artificial intelligence can be used. "The EU's AI Act is the first-ever comprehensive legal framework on Artificial Intelligence worldwide," claimed EU Commission president Ursula von der Leyen. According to …
I'm glad this 450 million-people block finally got to be first in something useful, after trailing in ExaFlopping HPC, and moon-and-mars landings (though they did quite well with that brain implant to make a paraplegic walk). The multi-tiered approach that they took seems quite appropriate, with separate considerations for "minimal risk" systems and for "high-risk" ones (like the immortal Kiss toy dolls for example: https://www.theregister.com/2023/12/04/kiss_are_now_immortal/ ).
However, you surely have a point that this EU Act may somewhat misidentify the target tech as AI, where it seems to mostly apply to Lossy Stochastic Databases. Accordingly, beyond the current provisional agreement, I think the bill should be renamed to: "The EU's LSD Act", which would be much more descriptive IMHO!
"I'm glad this 450 million-people block finally got to be first in something useful,"
You're an optimist, given that everything associated with Useless von der Leyen seems to end in messy failure. Hardly surprising as she's part of the European political aristocracy who cruises on her connections and family reputation despite the wreckage left in her wake.
Trust us, says EU, our AI Act will make AI trustworthy by banning the nasty ones
“Come into my parlour said the spider to the fly “ ....... springs immediately to mind, Laura, and is an APT* and quite appropriate reply to the result of the EU response to AI and the IT Sector as they spread their wings for future flights of fancy, systems takeovers and executive administrative makeovers
* Advanced Persistent Threat/Treat
manipulation of human behavior to circumvent free will
Seriously? You want to give lawyers a reason to argue about "free will"?
A problem that has beaten the finest philosophical minds of the past three centuries, and you're giving freaking lawyers a licence to debate it at $500 an hour?
Can't think who drafted that bit.
"manipulation of human behavior to circumvent free will
I think they're actually trying to do something about "dark patterns*" in UX here - you know, the "want to pay us 3 bones a month, forever? If you do not want this not to unhappen, please refrain from not unticking this box**" sort of business.
* or whatever it's called, I don't really remember right now
** apologies to Charlie Brooker for borrowing that
The free will remark jumped out at me, too. A friend know all your hot buttons, but doesn't push them. Ergo, if your buttons are being pushed, that's not your friend.
"dark patterns" -- That's Microsoft's entire UI these days. (Others are equally guilty.) I have never used Edge, so it has taken to opening full-screen after login. Or: "I had the MSFT store icon right there, but Steam isn't installed on this PC." https://forums.theregister.com/forum/all/2023/12/11/doom_30th_anniversary_sigil_wad/#c_4774074
Indeed. The press release page requires javascript to render at all -- completely blank otherwise. So I wasn't able to check whether the act actually uses the phrase "free will" (in translation of course) in respect of the "unacceptable risk" category. But the intent of the category seems clear and worthwhile. Let's hope it will stand up in the courts.
AI has jumped the queue. They were supposed to be clamping down on the metaverse before clamping down on AI. Simple because the metaverse just faded quietly away is no reason to ignore it. If you don't clamp down on things in the correct order, nobody will take you seriously.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
