Felix M. Simon

Should clarify that this isn't meant as a dig at Bengio whose work I respect. But I think statements like this illustrate how different parts of the "AI ecosystem" (for lack of a better word) don't properly engage.

Maybe it's more that despite this, nothing really seems to happen that puts a check on this power?

See for example a quick Google search across four outlets. Hardly representative, sure, but not exactly evidence of news not paying attention (and if you spend any time on BlueSky the discourse there is pretty lively on this issue).

While I look forward to reading the new International AI Safety Report, this response by Yoshua Bengio strikes me as a bit strange. It's not exactly that mainstream media aren't covering AI's power issues. Maybe it's more that despite this, nothing seems to happen that puts a check on this power?

Congratulations! Frankfurt is a great city (but then again I would say that :)

They know what they say is untrue, they do not care it is untrue.

The media know that what they repeat is untrue, they do not care it is untrue.

Their supporters know what they are told is untrue, they do not care it is untrue.

Telling the truth is not enough, unless people care about the truth.

I cannot tell you how absolutely infuriatingly this stuff comes across to Europeans.

When we say America is no longer a reliable ally, it goes considerably deeper than Donald Trump himself. This kind of complacent, condescending shit is a *huge* part of it.

Could Europe really leverage its $12.6tn pile of US assets? NIIPing an outlandish idea in the bud

"Every seller requires a buyer, and it’s unclear who would be willing to take the other side of any European fire sale of US assets."

www.ft.com/content/beea...

Nobel Taking The Piss Prize

It’s so useful

Nobel Please Prize

I hear #ica26 @icahdq.bsky.social notifications are going out, so here’s my annual reminder to put outcomes into perspective for early-career folks (full disclosure: I didn’t apply this year).

Interesting postdoc position on AI and journalism with @tomasdodds.bsky.social and the @publictechmedialab.bsky.social at the UW-Madison School of Journalism and Mass Communication. Tomas is a good egg, so do apply 🥚

wisconsin.wd1.myworkdayjobs.com/UW_Madison/j...

it's worth noting that the shift from Wikipedia to AI for factual summaries also removes our ability to audit the information trail and reveal systematic manipulation like this. AI can offer sources, but nothing like the timestamped/IP address-linked edit history of wikis
bsky.app/profile/timb...

About 18 months ago I sent this letter, along with a copy of my book, to a number of MPs and peers, including Wes Streeting. I never got a response (which to be fair was my baseline expectation), but it’s still disheartening to see today’s news about Haidt’s invitation to speak to policymakers.

The UK has some of the best social media and digital tech effects researchers in the world. I find it mind-boggling they haven’t been considered first.

I wonder if someone like @richardfletcher.bsky.social might be able to answer this based on @reutersinstitute.bsky.social survey data for the UK.

Partisanship is a helluva drug

✍How is GenAI affecting freelance journalism?
🔭@marinaadami.bsky.social wants to speak with:
1. Freelance journalists
2. Commissioning editors
about whether and how Gen AI is affecting their professions
📝Fill out this: buff.ly/Zgx6Z4R
📩Or get in touch directly at: marina.adami@politics.ox.ac.uk

Hahaha

@waqasejaz.bsky.social and colleagues

This is not my field and maybe they would actually recommend something similar at this point, but their output isn’t mainly op-eds…

Actual experts in the UK:

@orbenamy.bsky.social
@shuhbillskee.bsky.social
@sonialivingstone.bsky.social
@peteetchells.bsky.social

But no, let’s consult an academic activist who’s been called out on his methods more times than I can count.

And if you want an example of what Victoria sounds like, here is one of her podcasts: buff.ly/vdQhUfg

I haven't seen good evidence that this makes people stop believing everything they encounter (and I doubt it will), but we shouldn't ignore that it seems to be having a (small) effect).

Some of the research I cited from:

- buff.ly/bp0S8rS
- buff.ly/zE9Ve60
- buff.ly/a7ubs8Y

One additional point that didn't make it into the article but I think is worth noting is that this seems to be one instance of people becoming somewhat more sceptical of all media content they encounter, because they assume AI might have had a hand in it (or is wholesale responsible).

… (people listening to podcast because many like the fact that humans interact, not just the format of "audio-delivery") – and the challenges the use of AI brings with it in the journalistic context.

The piece Victoria has written is well-worth a read and looks at what AI voices can enable on the journalistic front (personalised podcasts anyone) but also where they might fall short…

Victoria hosts the FT's News Briefing podcast and once she put this issue to me over the phone I couldn't help but notice that I, too, see why some people thought she was AI: incredibly clear enunciation combined with her accent created something that many of the flagship AI voices resemble.