AI fails

Page may contain affiliate links. Please see terms for details.

Pblakeney

Shaman
Beware! The US may be coming for us all.
Although does accusing AI of being a failure mean you are a friend of SI? It gets confusing.

Screenshot 2026-10-08 at 22.42.27.png
 

Psamathe

Squire
Here is the link to the article in the Independent.

https://apple.news/AKM0bU1UsTnKf7z60Jxt2yA
I'm still sceptical.
I can't say they won't cause challenges but just 'cos AI can do one thing doesn't mean it can do anything.

Assuming it has resolved the Navier-Stokes equation the AI systems have also been working on tje other unsolved Millennium Prize Problems which they have not solved. Plus it. Seems that even for Navier-Stokes AI needed a lot of claimed help from academic mathematicians. One comment from a scientific podcast (New Scientist or Nature) was if you spent what OpenAI spent on solving it (US$15m) on human mathematicians you'd get a lot more out.
 
  • Like
Reactions: C R
OP
OP
briantrumpet

briantrumpet

Timewaster
Depressing, if not surprising to sceptics. Notice the correlation between writing skills and performance in science.

https://www.bennettschool.cam.ac.uk/blog/is-chatgpt-bad-for-students/

1791537809776.png
 

PurplePenguin

Über Member
Depressing, if not surprising to sceptics. Notice the correlation between writing skills and performance in science.

https://www.bennettschool.cam.ac.uk/blog/is-chatgpt-bad-for-students/

View attachment 17657

It's very helpful for education if used well. At the moment, it is probably used more as a mechanism to be lazy and to cheat; however, consider the question below which was written by a human (apparently). I can have a meaningful chat with Claude about this which is helpful. When I asked Claude to review the paper, it pulled up a number of errors including spelling mistakes in the actual questions. That might have been helpful for the author and improved the educational experience for the child.

1791538346086.png
 
OP
OP
briantrumpet

briantrumpet

Timewaster
It's very helpful for education if used well. At the moment, it is probably used more as a mechanism to be lazy and to cheat; however, consider the question below which was written by a human (apparently). I can have a meaningful chat with Claude about this which is helpful. When I asked Claude to review the paper, it pulled up a number of errors including spelling mistakes in the actual questions. That might have been helpful for the author and improved the educational experience for the child.

View attachment 17658

That's the thing... it could/should be a useful tool and of great benefit in specific contexts, but the AI companies want it be deployed universally in education and elsewhere, even though we *know* that if misused/overused will cause lifelong cognitive deficit.

Sceptics might suggest that a population-wide cognitive deficit might be an outcome that the AI companies are hoping for.

We managed to do spelling tests before the arrival of AI. In fact, I can remember a specific spelling test I did with Mr Lawes 51 years ago where one of the words tested was 'sceptic'. And, in any case, I'm pretty sure that there were spelling test programs/apps before AI/LLMs came along.
 

PurplePenguin

Über Member
That's the thing... it could/should be a useful tool and of great benefit in specific contexts, but the AI companies want it be deployed universally in education and elsewhere, even though we *know* that if misused/overused will cause lifelong cognitive deficit.

Sceptics might suggest that a population-wide cognitive deficit might be an outcome that the AI companies are hoping for.

We managed to do spelling tests before the arrival of AI. In fact, I can remember a specific spelling test I did with Mr Lawes 51 years ago where one of the words tested was 'sceptic'. And, in any case, I'm pretty sure that there were spelling test programs/apps before AI/LLMs came along.

I'm disappointed you haven't attempted to answer the question which is very much an old school question.
 

Pblakeney

Shaman
That's the thing... it could/should be a useful tool and of great benefit in specific contexts, but the AI companies want it be deployed universally in education and elsewhere, even though we *know* that if misused/overused will cause lifelong cognitive deficit.

Sceptics might suggest that a population-wide cognitive deficit might be an outcome that the AI companies are hoping for.

We managed to do spelling tests before the arrival of AI. In fact, I can remember a specific spelling test I did with Mr Lawes 51 years ago where one of the words tested was 'sceptic'. And, in any case, I'm pretty sure that there were spelling test programs/apps before AI/LLMs came along.

Consider me a sceptic. (I know that you already do)
Last thing these people want is an intelligent electorate in a democracy.
 
Top Bottom