I’ve seen some weird comments reacting to that negativity with "it has made some break through for science and medicine and blah blah..."

I was like "which ? Name one please.".

Maybe we should do the opposite and list the positive just so people realise how actually useful AI is.

September 02, 2024 - 13:06 UTC

3
0
3
This, for example... www.theguardian.com/technology/2022/jul/28/deepmind-uncovers-structure-of-200m-proteins-in-scientific-leap-forward
1
0
0
Prof Hannah Fry's podcast on this is very good. Unlike in the arts where I *hate* generative AI in particular, AI when applied to scientific problems can do amazing things... player.fm/series/google-deepmind-the-podcast
0
0
0
The issue there is that we call so many unrelated things AI. Like, the cancer detector thing? Super cool. It isn't in any way related to LLMs or art plagiarism machines, though, so its defenders unintentionally muddy the waters.
1
0
5
I’m not even sure it’s worth making the distinction. Machine learning has a lot of pitfalls (like any AI it’s a blackbox and some nasty false positive/negative are lurking) and often hardly beat a standard program written by skilled humans.

You are right... but it might not even help their case.
0
0
0
That famous "first image of a black hole" that made big headlines a while ago?

AI was used to aggregate the data and upres it.

There's plenty more, ML is used to parse data and quickly find patterns for scientists to verify in many areas, but that's one people remember.
2
0
0
The tech predates the current LLM and image gen gold rush (and moral panic) and it has plenty of legitimate uses. This "it's all bad and useless" trend is largely borrowed from the crypto panic and flies in the face of reality in key ways. Which is not to say there aren't big issues around those.
1
0
0
AI is essentially a pattern detection machine but it comes with a massive caveat, you have to reverse engineer the program to make sure it didn’t infer patterns from irrelevant or bad data.
1
0
0