[flagged]
ChatGPT n that are currently "disruptive". Disruptive to the point that I already have colleagues who get close to palpitations if they can't access it. Within mere weeks or a few months their entire modus operandi has changed. There are major issues (1) with LLM, not least their inscrutability and habit of hallucinating.
Soon we might be getting all whizzed up over "hyperdimensional computing" (2) which, let's face it has at least sorted out how to sound like a cool kid. However it is possible to work out how a hyperdimensional ... thingie "works" because of the way it is built and that is a powerful thing. ChatGPT n that are a bunch of weights and who knows what that actually means. No doubt we will have more stuff and frankenAI and who knows what else but it will be exciting.
I think things have only just started to get weird. It looks to me like a real paradigm shift.
To understand what the hell is happening we need to discuss things and that means discussing the old with the new - hence my comment. Feel free to disagree but please engage.
(1) https://arxiv.org/abs/2304.00612
(2) https://www.quantamagazine.org/a-new-approach-to-computation-reimagines-artificial-intelligence-20230413/
If the grandparent provided some analysis of what was output, maybe tied it back into the sorts of errors humans make and how gpt does or does not make dealing with this sort of issue easier, that'd be fine. I may or may not be terribly interested in it, but I wouldn't downvote it. But it didn't.
(edit to add: fwiw, I didn't flag the comment, and upvoted your reply)
The French Canadian bloke whose blog is linked here fed an exam to ChatGBT and it apparently passed and he mentioned others too. However my personal experience is rather less convincing. I have got it to write scripts and more but it lacks somewhat yet I still find myself saying please and thank you to it 8)
I'm not sure we even know what we don't know with respect to these things and that is a sure sign for me of a proper paradigm shift.