Will Lowe

@j2bryson.bsky.social

onlinelibrary.wiley.com

Experimental opera staging? Bring it on. But this interview suggests someone with no clear idea about what they were doing, or why.

Bayreuther KI-"Ring" startet: Marcus Lobbes über das Experiment | BR-Klassik br-klassik.de

Back from a conference with sessions on academic publishing and on how to fund the tokens for LLM-augmented research. Apparently there is no scheme where we give companies what we produce and they sell it back to us with extra noise & different fonts that we will not embrace. Some kind of fetish imo

I honestly wonder how much of the language policing around LLMs is just failure to appreciate how metonymy works. I just don't think that "Microsoft wants me to log in", "Brussels will capitulate" and "Claude said it's false" carry the ontological commitments their grammar might suggest.

The boring but effective way to ensure 'AI alignment' has two steps: 1. make those who build and sell AI responsible for what it does. 2. apply our best existing alignment technology – the law – to them. Maybe the AI is sentient, malign or benign. Who cares? It doesn't matter for this to work.

Here's a simple idea for seeing your way through the current 'AI in academic papers' panic. If you are the author on a paper, you're not claiming you did everything in the paper yourself (whatever that could even mean), you're taking responsibility for everything claimed in the paper yourself.

Why do I want access to all the books & articles? For training, testing & inference. But don't worry, it's 'transformative use'; what comes out will be very different from what I put in. It'll even grow the market for what I'm asking for. No, no, this isn't about LLMs. I'm talking about my students.

TFW you ask an LLM about how to derive the key result in a paper you've just read the abstract for, it seems to do so, but you can't check it because you can't access the paper. Well done to everyone involved in this farce.