Τεχνητή νοημοσύνη και η στάση του νομοθέτη

Αντιγράφω επιστολή αναγνώστη με τίτλο “Don’t regulate AI now”, η οποία δημοσιεύθηκε στην Εκόνομιστ της 11 Μαΐου 2023. Όπως θα περιμέναμε η παραδοσιακή αντίθεση ανάμεσα στην ρύθμιση και την ήπια διευθέτηση της αγοραίας διαδικασίας εκτείνεται (και) στην νέα περιοχή της τεχνητής νοημοσύνης.

“Your recent coverage on artificial intelligence was inspired, though your argument for regulating AI boiled down to comparing the technology to cars, planes and medicines (‘How to worry wisely about AI’, April 22nd). This is not a good analogy. Regulating AI is more like regulating physics or mechanical engineering, in other words it makes no sense. Specific applications of AI can be and are regulated, such as in self-driving cars. Many regulations assume human biases that are not present in AI (see chapter 21 of Daniel Kahneman’s ‘Thinking Fast and Slow’). The question is then whether these products should be regulated differently when AI is involved.

If AI gives rise to entirely new products then they should be regulated as such, if needed, but it would be quite premature to try to do so now. Regulating a nascent industry like AI opens the door to all sorts of noxious political and special-group interference. Imagine if the same had happened to the internet. The EU’s ‘regulate first, ask questions later’ approach is damaging to both consumers and the European tech industry. Kudos to Britain and America for being more sensible so far.”

Pedro Domingos
Professor emeritus, computer science and engineering
University of Washington
Seattle

Για την αντιγραφή

νκ

Share:

σχετικά άρθρα