2 Likes
it’s actually not new. the principles were invented in the 50s and 60s, died for a while because of the lack of processing and storage, and was “rediscovered” in the 80s and 90s.
vector processing became cheaper as mass production rose ( for consumer 3d graphics and then for crypto - re: nvidia is a 3 trillion dollar company now ), and warehouses of computers have become the norm for search, cloud services, and crypto.
that’s made this possible, but its limits are pretty well known. resource consumption is the backbone of making it work
imo the coolest ml has been for things like this:
so like cars or plastic, it can have it’s uses. but when it comes to these giant companies and their products: their profit is by definition our loss
and - unfortunately - that has to be addressed through regulation, because they have no motive to do it themselves
( eta: taxing the rich wouldn’t hurt either. a contributing factor is the untaxed money sloshing around that should be going to healthcare, food, and housing - but instead is funding a climate crisis )
4 Likes
I mean, it’s not like there isn’t enough mis- and disinformation and blatant inaccuracy out there in the world.
7 Likes
I’m largely pretty down on all the AI hype, but I’ve been finding more and more that for technical searches nothing really compares to ChatGPT. Using a regular search engine has results filled with click farm nonsense, shitty results, and half-baked Stack Overflows posts. Cutting out the middleman and having a generative AI craft results for me is often times so much more useful — especially for building code samples.
I only wish ChatGPT was better at citing its sources. This is something Bing’s AI is good at but much like Bing as a whole it just isn’t very good at technical searches and doesn’t support as many languages for generated code.
1 Like
Coming out of Boing Boing retirement to weight in here and let you know: you’re completely and utterly WRONG.
I have never had any problem getting great results on repairing computers using standard search terms and booleans. I built an entire career around it, and no one in any of the organizations I worked for was more respected than me despite knowing more. None. Today, I find AI tools to make it far harder and far WORSE to search for what I need. And that doesn’t even touch on the ethical and environmental concerns these tools engender.
Stop justifying your use of AI because your search abilities are weak. Google was my friend since 1998, and is no longer because they’ve forgotten what they were invented for: simple, easy, concise boolean search tools that WORK and don’t vomit garbage all over your screen while burning a train load of fuel for your query.
AI is a grifting pile of nonsense that’s wreaking havoc on the environment, creators, and other jobs. And most of it DOES NOT WORK. It’s just a way of a bunch of folks getting rich quick. Watch how many of them spin up an AI “solution,” profit from it, then sell it for massive amounts in just a couple of years. I guarantee it.
And go back and learn how to do proper searches and not rely on nonsense tech to help you.
12 Likes
I think the real trick is asking solid questions
That has ALWAYS been the trick with search engines. You’ve just replaced simple search terms (booleans) and going straight to wikipedia in the first place with far more complicated questions using an LLM that, as others who use these tools keep stating, often require further finessing and redirecting to get the right answers.
This is not an improvement. This was a solution to a non-existent problem.
EDIT: back into retirement for me. Stop using AI, kids. It’s not AI, and it’s ruining the world. Yes, researchers are having success with specialized LLM’s like MIT, but that’s not what we’re talking about here, and these Algorithmic Insanity (AI) tools are the latest in a long grift. Trust me, I’m dealing with all these fuckers at work as they push their fake AI into our organization and we’re fighting tooth and nail to keep people from rushing pell mell into these nonsense purchases.
Oh, and Boing Boing? Same goes to use, stop using AI. It’s cheap and degrades what you stand for. Pay real artists for the art on stories and support the people who made you what you once were.
12 Likes
It’s not about not knowing how to do proper searches, it’s that Google doesn’t work for proper searches anymore, as you well know.
As for A"I" search: yeah, it would be better if this had never happened, but the cat is out of the bag now. It’s a shiny and new cat that promises to make those that provide it to make them a lot of money, so it’s here to stay.
4 Likes
cfcs, asbestos, and coal mining made a decent amount of money. we had rivers burning in the 70s, and we’ve done pretty good on smog ( if not greenhouse gases themselves ) - just because something makes money, doesn’t mean we can reign it in.
part of the challenge is in getting people to realize that this isn’t all magically free
6 Likes
i guess that’s better than doing pretty bad?
my point is just that - contrary to expectation - we can reduce harm when we want to. we’re only mostly controlled by corporate interests; not entirely.
2 Likes
… I’m sure the air quality is much better in L.A. than it used to be, but this “fire season” the rest of us now have every year has got to be dragging the average down 
1 Like
Believe all you like, but we’ve all heard exactly that before, multiple times, from multiple blue chip companies and multiple Ivy League schools. And every time it has turned out to be marketing bullshit, hype, grift, and Potemkin AI.
Every. Fucking. Time.
6 Likes
ah the football of tech optimism.
this time for sure.
( eta: trying to make the world better by meeting people’s needs through public services… now that’ll never work /s )
6 Likes
So since this is apparently one of the most important examples, I’m curious…is that a lot? Like obviously two actual new antibiotics would be a big deal, but I know lots of things make it to testing and then die for being too toxic or otherwise impractical to actually use.
These two apparently didn’t kill the test mice, which is a good start, but I have no idea how remarkable it actually is or not to get a new drug candidate to that stage.
6 Likes
This topic was automatically closed after 5 days. New replies are no longer allowed.
