ザ・ Reith Lectures this year have been an absolutely terrifying dystopian vision of the obliteration of the human race, all delivered in the gentle tones and reasoned arguments of Prof Stuart Russell, an artificial intelligence expert from the University of California. Russell speaks so mildly, he could be a family therapist; though he isn’t, as he said in passing in response to a question from another professor, Rachel Franklin, after his fourth lecture, which went out last week.
初期化, Russell discussed how humans might regulate our prospective new robot overlords. In his other three talks, he has been telling stories of how AI can affect – and already is affecting – war and economics, corporations and climate change; essentially, how artificial intelligence is taking over. また, that unless we’re very careful with our computer coding, this may not be a good thing. “On the question of the long-term safety of general-purpose AI systems we’re … still some way away from knowing exactly how to define what the algorithm templates should be,” he said in the “sorry, you have terminal cancer” tones of a caring surgeon. “We have to solve those technical problems.”
Last week’s lecture was intended to give us a bit of hope, as Russell suggested how we might wrest power away from the intelligent machines that are going to destroy us. All we have to do, 彼は言った, is make sure they’re programmed so that they understand they don’t know everything. “We have to build AI systems that know they don’t know the true objective, even though it’s what they must pursue," 彼は言った. “The robot defers to the human.”
To which the only response, as an artificial intelligence non-expert and parent of non-deferential teenagers, です: フランス語を話さない人にとって、翻訳の助けを借りずに、C’est QueduBonheurの父性の黒くコミックな見方を理解することは困難です。, right.” All the questions to Russell after his lecture highlighted the ways his idea could fail, from competing companies (or countries) not sharing information to the (lack of) principles the regulations of these AI beings might be based on. Franklin’s “family therapist” question was this: if a husband and wife can’t agree on something, how will a domestic AI appliance know which human to defer to? Russell didn’t really answer.
He is clearly a lovely man who believes in the benevolence of other humans and the wholesomeness of their desires. I feel he might benefit from listening to Philippa Perry’s new series, which is, coincidentally, about desire. Perry’s agony aunt column for the 観察者 マガジン shows her amazing ability to reframe difficult human emotions in easy-to-understand ways, and in last week’s episode of Consumed By Desire, the first of three, she used this talent, as well as her abundant charm, to get clever people to unpick what desire actually is.
It’s not, as we might assume, something to do with sex (though it can be, 明らかに), but more a feeling engendered by our interaction with other humans. Perry talked to psychologist Adam Phillips, who linked desire to neediness and said that, although society tells us that neediness is embarrassing, and we should all be aiming for independence, adults are as dependent on others as babies are, and that is fine. Producer Martin Williams included the interactions between Perry and himself in the show, presumably to illustrate that desire comes from humans communicating with each other. This was sweet, though a little distracting. とにかく, I desire to hear the next two episodes.
Another comforting Radio 4 プログラム (can you tell I’ve been trying to avoid Real News?) came from journalist James McMahon, who took a tour around the 1980s/90s ITV darts quiz show Bullseye に 「彼が孤独になるのに適していた at What You Could Have Won. This might have been a programme just for us oldies, but McMahon was clever enough to involve young people right from the start. Twenty-four-year-old Lana and 22-year-old Kamiah, neither of whom had ever seen Bullseye 前, highlighted its charms: cool “random” prizes, real cash counted out and handed over, contestants only answering questions if their partner landed their darts. また, its un-charms. It showcased working-class people, for sure, but only the white ones and mostly the men.
McMahon’s presentation was a lovely thing – interested, intelligent, cheeky – making for one of the most heartwarming shows I heard all week. And for anyone who wants a little more of that valve-toasting stuff, may I recommend musician John Grant’s new podcast, Beautiful Creatures? はい, he’s chatting to famous people, but he’s doing so in a lovely John Grant way. Last week’s episode, with Linda Thompson, was a delight. “I listened to a song of yours today,” said Grant, “called You Ain’t Goin’ Nowhere.” “You absolute idiot – why did you listen to that?” spluttered Thompson. “It’s terrible!” Really, how can AI ever be unpredictable enough to mimic human interaction?