What I dislike about AI is the unthinking Kludge science which underlies it. LLM’s are just data mining. Combining several sources of info is bookkeeping, not science.
AFAIK there is nothing one would call " intelligence" under the bonnet.
It works because of the huge scale of its training data… but at some cost… it takes a nuclear reactor to power one AI data centre.
I want to see something that is more based on our understanding of how things work… ie base it on the ‘laws’ of science, not on observations.
Moving on further into this subject, I have just come across a very interesting (but long) YT video by Jay LaCroix of LearnLinuxTV fame where (inter alia) he talks about the downsides of AI 44:45m in). But the really amusing part is at ~ 42m where he asked AI to respond to some outrageous questions with the responses it provided.
I don’t use AI. Period. No interest. I tried a bit of chatgpt (I think?) and was just not impressed. “I can do the same thing just by searching with Kagi.” We’re encouraged to use it responsibly at my job. I am not convinced this drivel is of any use. I even asked for a way to disable M365 CoPilot, as I flatly stated in an “AI survey” that it had no place at the dinner table. I even turn off AI summaries (and have turned off AI slop in Firefox).
Sorry. I know this sounds like a rant, and I guess it is. I simply don’t see the benefit of it. Not in its current form. So, to answer the initial query of this topic, it should be a requirement to divulge “I use AI garbage in my work. That way we know it’s not really you.”
There used to be a hotel downtown where Grant Wood created a mural. It was famous, but it was torn down decades ago. I thought I remembered the name of the hotel, but I thought I’d ask Copilot on my phone. It confirmed I had the correct hotel name, The Hotel Montrose. https://www.thegazette.com/news/time-machine-the-hotel-montrose/. But then it volunteered more information that was interesting, but incorrect. It gave the wrong address.
I replied to it that I thought it was at the correct address. Of course, it replied something like, “You’re absolutely right question that…” Then it confirmed I was correct on the location and proceeded to, again, give more information.
I followed that up with another question, because I knew the answer. I asked what was at the former Hotel location now? It said it was the Kubias Building.
I corrected it and said the Kubias Building was a block east of there. Hibu, where I work, is at the former Hotel location. It came back with, “You’re absolutely right…”. Then it did seem to know exactly what was there.
I thought it was kind of funny. You can’t take the responses on faith; you need to verify.
On the software development front, I have been using GitHub Copilot and Amazon Q to do some home development. It’s pretty astounding how good it can be. It isn’t perfect, but it speeds development (for me) 300%. Just throwing out a number, but it’s huge. You do need to be careful. We had a Hackathon at work recently and the code produced by Cursor (in this case) did a bit more than we intended. We headed it off before it affected production servers. Sheesh.
The Ontario government is dipping its toe into the rapidly growing world of AI, telling civil servants they can use Microsoft Copilot in their day-to-day jobs and offering some more advanced tools.
The move appears to have been embraced, according to internal metrics, which show the Ontario Public Service has the highest Copilot chat use in the country.
Government presentations, obtained using freedom of information laws, show more than 15,000 civil servants use the artificial intelligence assistant every week, with more than 120,000 pageviews for the “Copilot Chat InsideOPS” page.
Copy and paste is about the same. LOL AI is everywhere
That is quite worrying, though the government was about collecting taxes, empty the bins, provide housing and schools et. A lot of time wasting springs to mind !
Well I would say it is ‘stolen’ rather than ‘owned’.
You cant own something that you copied from the public domain, no matter how you rearrange it.
The whole concept of ‘ownership’ is a sad reflection on the materialistic nature of our world.
The best way to ‘own’ something is to give it away… free, and free of restrictions… like free software. GNU still ‘owns’ their unix utilities, in a much better way than they would have if they had commercialised it. Linus still ‘owns’ Linux.
When are the AI providers going to fully embrace the ‘free content’ movement?