i fucking hate these people so much man
im getting fired if i dont use this bullshit, already one person was “anonymously” shat on during a meeting for not doing so. fortunately not me
i fucking hate these people so much man
im getting fired if i dont use this bullshit, already one person was “anonymously” shat on during a meeting for not doing so. fortunately not me
LLM? Not so much IMO. But neural nets and deep learning are genuinely cool and useful for things like image recognition, speech recognition, etc. As an aide to a doctor reviewing a colonoscopy scope, sure. But instead they will treat it as a replacement that the doctor rubber stamps. Due to the way we will use it, they are dangerous, even if the tech itself can be helpful.
Theres still the issue of electricity, of the human and environmental costs, etc.
A good metric: if correctness is unimportant and verification is easy, then it is not an inappropriate tool (that doesnt mean its necessarily a good tool for a given task though). However if correctness is important and verification is difficult, then it is not appropriate. Basically, dont use it to summarize that many-page report, because its nondeterministic and you’ll have to read the whole thing anyway to see if the LLM actually told you all relevant portions and didn’t generate things that aren’t in the report. If you want a summary, thats literally what the abstract and conclusion parts of a report are for. Dont use it to write fresh code for you, cause you’ll have to read through all of it, build a mental model, and understand the problem space inside and out to actually review it well, and at that point you might as well have written the code.
I think that’s one of the things that irks me most about this LLM bubble. I think ML has really cool applications in research, approximate tasks with margins for error like image classification, upscaling etc. It was one of my favorite topics during my CS degree. But now it’s tainted by the gen “AI” slop machines, so all the interesting breakthroughs of real practical ML get chalked up to “AI” innovation and people just assume chatGPT is responsible when it’s actually a team of researchers building small efficient programs. Ofc ML is less efficient than traditional computational approaches that can work deterministically so those should always be prioritized, but it has its uses for sure.
This right here. The chat bot shit is shit. The tech analyzing X Rays to detect lung cancers is great. The problem is they can’t just sell “machine augmented professionals” they need to sell “A.I.” which is a marketing department fever dream
well, it’s great if it’s actually doing something novel and not just noticing that there’s a strong correlation with older medical equipment, certain kinds of poverty, and smoking, so it’s just assuming a shitty xray is cancer.
Also your radiologists who use it can become deskilled because they rely too much on the AI, which can cause problems over time
My favourite go-to example is the DARPA gun that learnt overcast day = hidden tank
https://gwern.net/tank
(Though I’m arguing against myself because the above link says it’s an unsourced, possibly overblown urban legend, idrc)
There’s a few problems (like image recognition) where deterministic approaches are either nonfunctional or far more resource and time intensive. But like, thats not what we get. Capital declares we shall feed it so we feed it. We will hit an ai winter again. Third time and its gonna hurt more than the others. We had one in the 70s, one in the 90s (though that was also affected by the dismantling of the soviet onion and drying up of DOD funding), and now its wormed its way into everything and is gonna be painful when it hits.
The various projects I say pre-2020 involving machine learning were genuinely interesting.
You were seeing a computer think in a way analogous to biological brains.
And then the LLMs came.