Speaking as a researcher in the AI/ML space, you see an interesting divergence at the upper end of familiarity. Some focus on the rapid recent growth of the technology and are quite enthusiastic about it, but others (myself included) focus much more on its limitations, especially driven by the type/quality of the training data used to make the models. I think it comes from different backgrounds prior to learning the tech. Computer science people tend to be in group one, whereas other scientists (biologists, physicists…) that adopt ML as a tool are more likely to be in group two. To be clear, all of this is my personal experience from personal interactions and literature review in graduate school, not some large scale survey.
Comment on Knowing less about AI makes people more open to having it in their lives – new research
thebardingreen@lemmy.starlightkel.xyz 4 weeks ago
Interesting. I bet it’s more like this.
bananaslug4@lemmy.blahaj.zone 4 weeks ago
remotelove@lemmy.ca 4 weeks ago
Maybe. People with more technical knowledge should understand that LLMs aren’t magic or sentient and have some severe limitations. Hell, I have been tinkering with ML and ANNs for a better part of 15 years or so and they can be extremely useful. (I am no expert and never indend to be.)
It’s the marketing wank, scams, art theft and all the bullshit that pisses me off now. In that regard, I am squarely in the “Fuck AI” category. There is absolutely nothing phenomenal that has come of this recent bubble in the commercial space. AI generated images are mostly trash, articles are riddled with gross factual errors phishing and other scams are more realistic (and maybe even more dynamic) now and public forums contain even more annoying bots. And the worst bit is that AI generated media, like music, is just a collection of averaged values with no originality.
That bell curve represents something but it isn’t IQ.
termus@beehaw.org 4 weeks ago
My main thing is I hate it being used as a replacement for human creativity. I definitely think it has its uses. I like using DLSS and DLAA in games that support it. For the most part it’s made my gaming experience better. I see it’s usefulness with generating conversations for NPCs in a game. That’s crazy cool but it needs to be tiptoed around. It shouldn’t be used to replace writers, voice actors or artists.
I signed up for Midjourney a while back to see what all the hoopla is about. As I sat there refining my prompt and regenerating OVER and over and over to hopefully get something that wasn’t all fucked was such a waste of my time and an exponential waste of compute and electricity to generate what is essentially instant garbage. It’s crazy and I felt bad about using it afterwards.
Bronzebeard@lemm.ee 4 weeks ago
It’s probably that people familiar with the technicals of AI don’t automatically assume it’s LLMs being talked about. There’s several other types being out to very good use, especially in the sciences right now.
Even LLMs have their place as interpretors for those other types. Having natural language interpretation that is fairly accurate the vast majority of the time is a powerful tool with a lot of applications. Near Real time translation or closed captioning is great for accessibility.
Free_Opinions@feddit.uk 4 weeks ago
One major issue with most online discussion about AI is that many people seem to view that term synonymous with LLMs - which it’s not. LLMs are AI systems but AI as a category is also so much more than that.