You are given an urn containing 100 balls; n of them are red, and 100-n are green, where n is chosen uniformly at random in [0, 100]. You take a random ball out of the urn—it’s red—and discard it. The next ball you pick (out of the 99 remaining) is:

While GenAI is fun, I think its economic value is grossly over estimated, because it’s unreliable, risky and expensive to make and serve. It’s fine for creative tasks, but not (yet) autonomous agents

Sigmund Freud on lecturing, 1884

Jasmin Paris Jasmin Paris finished loop five of the #BM100 in 59:58:21.

If you started with a large budget and a goal of reducing traffic fatalities, what would you do? I don't think autonomous vehicles (of the self-driving car flavour) would feature for me, given the proven alternatives for getting people from A to B safely.

The tedious stuff requires accuracy.

There are sensible AI use cases, and then there are things like thus. Anything that removes humanity from the Olympics will make it worse. theguardian.com/sport/2024/apr…

The personals section in the latest London Review of Books has got something for everyone

When AI people talk about labour, they tend to frame jobs as sets of tasks, rather than sets of responsibilities.

Nope

Really looking forward to this from Neil Lawrence - penguin.co.uk/books/455130/t…

They've really caught some big ones with this post 🎣

Pyrotechnic football writing on deadline by Jonathan Liew in trying to describe how badly 🏴󠁧󠁢󠁥󠁮󠁧󠁿 played tonight

An article that features in Neil Lawrence 's great book IIRC.

Hey Edinburgh Festival Fringe! Do you realise Anthropic AI rips off your performers past and present? Just a quick look at Alex Reisner's new tool shows work by Stephen Fry, Jack Whitehall, Graham Norton, Darren McGarvey scraped from YouTube without consent or remuneration to train Claude.

It's been slowly dawning on me that the claims of AI reaching superhuman levels from the tech bros has more to do with their horribly narrow definition of human-ness than any improvements in their AI. And @ShannonVallor nails it here.

Cool - some of the boats doing this are tiny! theguardian.com/environment/ar…

Quite a lot of my feed on her be "I made this app in <insert framework> just using <insert language model> without needing to understand!". Imagine a bad actor creating /fintetuning an LLM that injects security vulnerabilities. Absolute carnage.

Three articles in the London Review of Books this month that taken together are quite chilling: the wholesale swallowing of the religion of AI (article on Blair) + the use of AI in generating kill lists + the hollowing of the regulatory landscape (in context of grenfell)...

Enjoying this podcast series featuring Melanie Mitchell on intelligence -- highly recommended santafe.edu/culture/podcas…