Light and shade
In College the other day.
Quote of the Day
”Commerce in the 21st century is espionage for profit.”
- Historian Jill Lepore, writing in the New Yorker.
Musical alternative to the morning’s radio news
Ray Charles | Georgia On My Mind
Georgia may also be on Trump’s mind, but for a different reason.
Long Read of the Day
AI Chatbots Don’t Care About Your Social Norms
They seem to fool people into thinking they’re human, argue Jacob Browning and Yann Lecun in this essay but they are actually exceedingly alien.
With artificial intelligence now powering Microsoft’s Bing and Google’s Bard search engines, brilliant and clever conversational AI is at our fingertips. But there have been many uncanny moments — including casually delivered disturbing comments like calling a reporter ugly, declaring love for strangers or rattling off plans for taking over the world.
To make sense of these bizarre moments, it’s helpful to start by thinking about the phenomenon of saying the wrong thing. Humans are usually very good at avoiding spoken mistakes, gaffes and faux pas. Chatbots, by contrast, screw up a lot. Understanding why humans excel at this clarifies when and why we trust each other — and why current chatbots can’t be trusted.
Getting It Wrong
For GPT-3, there is only one way to say the wrong thing: By making a statistically unlikely response to whatever the last few words were. Its understanding of context, situation and appropriateness concerns only what can be derived from the user’s prompt. For ChatGPT, this is modified slightly in a novel and interesting way. In addition to saying something statistically likely, the model’s responses are also reinforced by human evaluators: The system outputs a response, and human evaluators either reinforce it as a good one or not (a grueling, traumatizing process for the evaluators). The upshot is a system that is not just saying something plausible, but also (ideally) something a human would judge to be appropriate — if not the right thing, at least not offensive…
Great essay. The authors’ conclusion ought to be printed in 95-point Helevetica Bold on every schoolroom and lecture-hall wall:
The upshot is that chatbots aren’t conversing in a human way, and they’ll never get there solely by saying statistically likely things. Without a genuine understanding of the social world, these systems are just idle chatterboxes — no matter how witty or eloquent.
I’ve written for the Observer for a long time (my first piece in the paper was published in 1972, I think, and I’ve been a weekly columnist there ever since 1987).
At one stage during that time, the paper had my fellow-countryman Conor Cruise O’Brien as its Editor-in-Chief. He was a big figure in every sense of the word — an experienced diplomat and UN official (he had been the UN’s High Representative in Katanga during the Congolese civil war), the Schweitzer Professor of the Humanities at New York university, a Cabinet minister in Ireland, a distinguished author of fine books (on Camus and Edmund Burke, to name just two), and a major public intellectual.
He was also a formidable drinker, and if you went to the pub with him on a Friday night you needed to be ready for anything. One evening he said to me, “Is it true that you’re an academic as well as a journalist?”. “I’m afraid it is, Conor”, I replied. “I see,” he said. “Same as me: you have a foot in both graves.”
He liked the Observer but I think he regarded many of us as woolly-headed liberals. Still, he appreciated the ethos of the paper and occasionally told stories against himself about it.
One was about a day when he was phoning in the copy for his column from some distant land and at one point dictated as follows to the copy-taker: “the atmosphere was redolent of fin-de-siecle Vienna — that’s French – f-i—n-space-d-e-space…” At which point the copy-taker politely stopped him and said, “I think you should take it for granted, Dr O’Brien, that a copy-taker on the Observer would know what the French for ‘end of the century’ is.”
It was a good story and we all laughed at it. But although I knew the phrase at the heart of the story, I had no real understanding of what it implied.
I’m reading — for the first time — Stefan Zweig’s The World of Yesterday, and his account of the Vienna in which he was a precocious teenager and young man is one a truly remarkable evocation of a special era. I’m finding the book unputdownable (to resort to cliché) and am retrospectively grateful to Clive James whose essay on Zweig in his collection Cultural Amnesia was what started me down this enjoyable rabbit-hole.
This Blog is also available as a daily email. If you think that might suit you better, why not subscribe? One email a day, Monday through Friday, delivered to your inbox. It’s free, and you can always unsubscribe if you conclude your inbox is full enough already!