AI fails

[briantrumpet]

Prize to anyone who can make baked beans do that.

 

[Psamathe]

Whilst air flights don't seem to be lowering their climate impact, maybe the bigger issue is our continually having to ask AI trivia

Drax datacentre could create nearly twice the yearly emissions of all Gatwick flights​

Drax’s planned datacentre could produce nearly double the emissions of all flights out of Gatwick airport each year by burning 4.9m tonnes of wood annually, an analysis has shown.

 

[rakkor]

Why do all the techbros sound the same - https://www.theguardian.com/media/2026/oct/06/openai-australia-parliament-inquiry-jason-kwon

 

[LCpl Boiled Egg]

Prize to anyone who can make baked beans do that.

View attachment 17610

Looks like a jacket spud with Maltesers and Chipsticks to me. I'd give it a go to be honest. I've eaten worse.

 

[briantrumpet]

Not surprisingly, Alex Harvey (human) is more than a little pissed off by Amazon and their 'Andrew V. Smith' AI creation.

 

[briantrumpet]

Not surprisingly, Alex Harvey (human) is more than a little pissed off by Amazon and their 'Andrew V. Smith' AI creation.

View attachment 17638

Seems that 'Andrew V. Smith' has 'written' 31 books in the past year.

 

[briantrumpet]

I'd say, "Parenting works different when your brain has been fried by AI".

 

[briantrumpet]

Haha. And then he thinks this makes it better. Maybe his wife could replace him with a turkey baster if she wants any more children.

 

[Pblakeney]

Haha. And then he thinks this makes it better. Maybe his wife could replace him with a turkey baster if she wants any more children.

View attachment 17644

I suppose I'd be wasting my time trying to convince him that from the child's perspective that time spent reading stories is very valuable, precious, and not wasted time in any way or form. I think it says a lot about him. Yeah, use the baster instead of the waster.

 

[briantrumpet]

I suppose I'd be wasting my time trying to convince him that from the child's perspective that time spent reading stories is very valuable, precious, and not wasted time in any way or form. I think it says a lot about him. Yeah, use the baster instead of the waster.

Reading that second one again. I think he'd be better off with a small dog than a small person. Seriously. He's been entrusted with one of the wonders of the universe (a child's brain), and all he wants to do is use the child as a comfort toy for himself.

Mind you, not appreciating, caring for (or even understanding) the human brain seems to be a feature of AI bros.

 

[briantrumpet]

This also collapses into a comprehension fail on my part, but there's something about turning words about maths into computational language there, I think. And Navier-Stokes is referenced, the thing they claimed AI had solved.

https://arxiv.org/abs/2610.08144

Autoformalisation is increasingly used to verify mathematical texts, including those generated by AI, as in OpenAI's announced proof of blow-up of solutions to the Navier-Stokes equations. In this process, an AI system translates the text from a natural language (NL) into a formal language such as Lean. Once this translation is done, the argument expressed in the formal language can easily be mechanically verified. The purpose of this article is to demonstrate why this process may offer no confidence in the original NL argument, owing to the various difficulties in performing the translation semantically faithfully. In particular, we highlight that the problem of resolving ambiguities in mathematical NL text, which is necessary in order to provide semantically faithful translation, is arbitrarily high up in the Solvability Complexity Index (SCI) hierarchy/arithmetical hierarchy (the SCI =∞). Hence, informally, providing semantically faithful AI autoformalisation is harder than any computational problem including the Halting problem (which has SCI =1). To demonstrate the effect of this result we provide several examples of AI mistranslations of NL statements and proofs into Lean in practice, resulting in mismatches between NL proofs and their Lean `verifications'. These include OpenAI's announced Navier-Stokes proof. In particular, we show that the formalised Lean proof does not correspond to the NL proof of blow-up of solutions to the Navier-Stokes equations.

 

[Beebo]

Just read a news report that AI super mathematics will soon be powerful enough to break down all encryption on the net. This includes the blockchain which holds up all cryptocurrencies.

 

[Psamathe]

Just read a news report that AI super mathematics will soon be powerful enough to break down all encryption on the net. This includes the blockchain which holds up all cryptocurrencies.

I'm sceptical. Factoring large numbers is a difficult for an AI as it is for any software based on conventional processor technology. Might be these vast data centres are so stuffed full of processors it becomes faster but at significant expense.

The risk to widespread encryption comes from Quantum computers but even then there are steps to safeguard using structured lattices and hash maths which is hard for quantum computers as well ss conventional processors) to solve (post quantum encryption).

 

[Pblakeney]

I'm sceptical. Factoring large numbers is a difficult for an AI as it is for any software based on conventional processor technology. Might be these vast data centres are so stuffed full of processors it becomes faster but at significant expense.

The risk to widespread encryption comes from Quantum computers but even then there are steps to safeguard using structured lattices and hash maths which is hard for quantum computers as well ss conventional processors) to solve (post quantum encryption).

This reads like the terrorists handbook.
Terrorists don't need all missions to be successful, just one will do the damage. Sometimes just the credible threat is enough.

 

[Beebo]

I'm sceptical. Factoring large numbers is a difficult for an AI as it is for any software based on conventional processor technology. Might be these vast data centres are so stuffed full of processors it becomes faster but at significant expense.

The risk to widespread encryption comes from Quantum computers but even then there are steps to safeguard using structured lattices and hash maths which is hard for quantum computers as well ss conventional processors) to solve (post quantum encryption).

Here is the link to the article in the Independent.

https://apple.news/AKM0bU1UsTnKf7z60Jxt2yA