AI fails

[C R]

This is roughly where I think it's probably at... AI has been mahoosively oversold on the back of its LLM output (which is extraordinary, despite its current shortcomings). The poster here is a tech writer at the WSJ, not a nobody on a cycling forum.

View attachment 15654

I am not sure I completely agree with the third one.

 

[AndyRM]

I am not sure I completely agree with the third one.

I suppose it will, but not in a good way...

 

[C R]

I suppose it will, but not in a good way...

My experience is mixed. Overall the net time saving is zero, some tasks it does quickly and right, others it needs so much prompting that I would have done a better job quicker myself. There's still a bit of an emperor new clothes thing with people not wanting to acknowledge some obvious shortcomings.

 

[briantrumpet]

I suppose it will, but not in a good way...

It'll be bad if coding gets to a point when real humans don't/can't understand how AI is developing it. Such is the short-sightedness of the industry that I could see it heading in that direction.

 

[briantrumpet]

Doesn't sound great.

https://www.chronicle.com/article/my-students-cant-read

 

[briantrumpet]

Doesn't sound great.

View attachment 15666

https://www.chronicle.com/article/my-students-cant-read

Here's the cheeky access to the article.

https://archive.ph/WvW1F

 

[briantrumpet]

Here's the cheeky access to the article.

https://archive.ph/WvW1F

Blimey, that's really bleak.

 

[PurplePenguin]

Here's the cheeky access to the article.

https://archive.ph/WvW1F

I see their game. Write a long article about how people don't read long articles any more.

 

[briantrumpet]

I see their game. Write a long article about how people don't read long articles any more.

Haha, the thought did pass through my prehistoric brain too, though I'd not really class that as a long article. Anyway, maybe they should turn it into a 20-minute TEDx video, or a podcast where they read out tweets slowly.

 

[Psamathe]

Interesting aspect to what can happen when an AI "fails"

Doctors and NHS could be sued for mistakes made by AI tools, report warns
Doctors and the NHS could be sued for medical negligence over mistakes made by artificial intelligence tools used in diagnosing patients and suggesting their treatment, ministers are being warned.

Under the law as it stands, medics and the health service can be held liable for patients being harmed or dying even if it was AI that made the errors that resulted in their suffering.

 

[PurplePenguin]

AI is good for a chat and is very good at validating your feelings. That's not a fail.

 

[AndyRM]

My good lady discovered yesterday that you can ask Gemini to do a quiz while you're on the go, which is very handy for long car trips.

For the first time ever I have approved of something done by AI.

 

[Pross]

Did it give correct answers though?

 

[Ian H]

AI is good for a chat and is very good at validating your feelings. That's not a fail.

The only AI chat I have undertaken was an option at the end of a YouGov survey. I found it singularly unpleasant.

 

[AndyRM]

Did it give correct answers though?

Yup.

And even had a bit of humour about it too, after it repeated a question and was told off, started reading the same question again before saying "I bet you thought I was going to repeat myself..."

A bit naff, but fun.