Tobacco has caused great harm for little or no benefit, and I’m not sure it has gone away (globally) yet. Anyhow, if the benefit is to the 1% (or 0.1%) the lack of benefit to the rest of us is not going to matter.
I agree that people will eventually get bored of chatbots, and will spend less time on them. But using AI for coding is not going away. I’m judging humor books for a contest, and two of them have AI-generated illustrations. (Not text, so it is okay.) That’s bad for illustrators, though it is not clear if the authors could have afforded an illustrator. That’s not going away. I suspect AI-generated porn is not going away.
A crash from overbuilding is pretty much inevitable. But that’s not the end of the world.
The 1% won’t get a choice when governments decide to metaphorically or literally put a gun to their heads. Not every nation is the US. Not to mentioned they won’t be the “1%” anymore if they cause a general economic collapse.
And it won’t be used for coding for all that long, given the buggy garbage it produces. Either companies will stop using it willingly, the governments will step in and stop them, or our technological infrastructure will collapse under the bugs and there won’t be any sort of coding at all going on.
Any such AI will have no meaningful relationship with the present ao0called “AI” chatbots, which are primarily a vehicle for scam.
And there won’t be any “evolving merger of human intelligence with machine intelligence”; if AI gets that good the wealthy and powerful will just have the machines kill the rest of us and live in their own little robot-run palaces.
You’re assuming chatbots are the end goal of AI. They are just a step in the process. A decade ago the best AI programs were playing atari games and learning to play go and starcraft. But nobody thought that was the end goal of AI, just a step as it learns to do better things.
Right now AI is good at sifting through existing knowledge to find answers. But newer architecture will be better at discovering new knowledge and adding that to humanities knowledge base.
Programs like alphafold, GNoME, etc are going to get more advanced. Chatbots are a small part of what AI can do and will do.
AI is already starting to show modest improvements in worker productivity.
No; they are what people are calling “AI”. Real AI will have little if anything to do with them. They are an unsustainably destructive dead end.
I think that train has left the station. I’ve read enough from programmers who have used AI about how it found bugs in code and did a pretty good job quickly. A lot of code is pretty trivial.
I’ve just quit from working on a big conference, and one of the things I had was some code to convert tables from the system we used to collect papers into a Word document for the schedule. I turned over my code, which is Perl and Python, and one of the committee members who is a researcher at IBM said she is going to use AI to redo it. We’ll see how it works out, but I bet AI can do this. Most code is not super complicated. Finding the corner cases and making sure the requirements make sense is where most of the problems come from.
I took my first AI class in 1971. There has been remarkably little work on “real AI” since then, though all the stuff we studied back then as research now sits in our phones.
I have this suspicion that AI will not achieve some heights of real intelligence, but show us that much of what we consider our special intelligence is no big deal. Back then some people still thought that chess was something that required real intelligence to do well.
As opposed to giant, energy and water guzzling “data centers”. This so-called AI is a dead end in part because civilization cannot sustain it. It does a worse job than a human you can power with a hamburger.
That’s different. A good chunk of those will either be growing weeds or turned into Amazon fulfillment centers or never get built at all. Right now we have 20 companies each thinking they’ll be getting 40% of an inflated market. And sucker governments giving tax breaks, not realizing that after the construction is done hardly anyone will be working in these places.
It is full Florida land boom. 60 years ago my father got suckered into buying land in Rio Ranchos, NM. The address is the corner of cactus and cactus. I own it now, and the property taxes are very low, But I’ve just gotten some reasonable offers to buy it. I don’t need the money, and my son-in-law would like it some day, so I’m not selling. I find no evidence that a datacenter will be built there, but someone is speculating.
I was wondering…these "new"AI assistants are being advertised to do all sorts of personal tasks, clean up your files, make schedules etc etc.
My question is this: with all the talk of bots flooding much of the internet, and the ability of for instance, Chat gpt to argue quite well, (annoyingly well)…how can we ever tell if say, a new member here, or any forum, was not an AI programmed to interact as “humanly” as possible? That just seems so trivially easy?
" go to my account at the Straight Dope and finish my argument about airplanes on treadmills"
Has there ever been one here at the Dope?
Trick question: The Straight Dope Message Board does not get new members any more.