I am thinking of deploying a RAG system to ingest all of Linus’s emails, commit messages and pull request comments, and we will have a Linus chatbot.
Hold on there Satan… let’s be reasonable here.
The only time I’ve seen AI work well are for things like game development, mainly the upscaling of textures and filling in missing frames of older games so they can run at higher frames without being choppy. Maybe even have applications for getting more voice acting done… If the SAG and Silicon Valley can find an arrangement for that that works out well for both parties…
If not for that I’d say 10% reality was being… incredibly favorable to the tech bros
^^
^^^
I play around with the paid version of chatgpt and I still don’t have any practical use for it. it’s just a toy at this point.
I use shell_gpt with OpenAI api key so that I don’t have to pay a monthly fee for their web interface which is way too expensive. I topped up my account with 5$ back in March and I still haven’t use it up. It is OK for getting info about very well established info where doing a web search would be more exhausting than asking chatgpt. But every time I try something more esoteric it will make up shit, like non existent options for CLI tools
ugh hallucinating commands is such a pain
I used chatGPT to help make looking up some syntax on a niche scripting language over the weekend to speed up the time I spent working so I could get back to the weekend.
Then, yesterday, I spent time talking to a colleague who was familiar with the language to find the real syntax because chatGPT just made shit up and doesn’t seem to have been accurate about any of the details I asked about.
Though it did help me realize that this whole time when I thought I was frying things, I was often actually steaming them, so I guess it balances out a bit?
It’s useful for my firmware development, but it’s a tool like any other. Pros and cons.
Nice replacement topic after the maintainer drama last week
I think the drama came from when the Russian forces started killing civilians 🤷
Not a company following the law.
Sucks to
suckwork for companies run by a wartime government.Yea this is so blatant I’m not even going to click on that shit.
So basically just like linux. Except linux has no marketing…So 10% reality, and 90% uhhhhhhhhhh…
What
Some Linux bad Windows good troll
Did I fall into a 1999 Slashdot comment section somehow?
So basically just like linux. Except linux has no marketing
Except for the most popular OS on the Internet, of course.
90% angry nerds fighting each other over what answer is “right”
That says more about your ignorance than anything about AI or Linux.
You’re aware Linux basically runs the Internet, right?
You’re aware Linux basically runs the
InternetWorld, right?Billions of devices run Linux. It is an amazing feat!
Never heard of Android I guess?
What happened to Linus? He looks so old now…
If you find out what happened, let me know, because I think it’s happening to me too.
He got old.
Not especially old, though; he looks like a 54yo dev. Reminds me of my uncles when they were 54yo devs.
As a 46 year old dev I’m starting to look that way too.
I guess having 3 kids will do that to you.
That, and developing software for 30+ years.
That and leading an open source project for 30 years.
THE open source project.
Whether you’re leading a project or not, time will have pretty much the same impact. He’s in his mid-50s, and he looks pretty good for that age.
I mean he’s aging quite well given his position… Many people burn out way earlier.
[citation needed]/s
That’s an excessive amount of aging is what folks are seeing. Not that he’s just old.
He’s lost a lot of weight in 4 years so that’s probably exacerbating the wtf.
He’s 54, I think he looks pretty average for that age. He looks like an old dad, because he is.
I told him not to go to that beach.
Oxidative stress is a bitch
he aged
Source?
What happened to he is happening now to you.
Wow, yeah that’s a big difference from how I remember him
He’s 54 years old
People age. You don’t look the same as in 2010 either, I know that without having any idea what you look like.
He has a real Michael McKean vibe
Time
It’s like he aged 10 years in the past 2 years… damn
Honestly, he’s wrong though.
I know tons of full stack developers who use AI to GREATLY speed up their workflow. I’ve used AI image generators to put something I wanted into the concept stage before I paid an artist to do the work with the revisions I wanted that I couldn’t get AI to produce properly.
And first and foremost, they’re a great use in surfacing information that is discussed and available, but might be buried with no SEO behind it to surface it. They are terrible at deducing things themselves, because they can’t ‘think’, or coming up with solutions that others haven’t already - but so long as people are aware of those limitations, then they’re a pretty good tool to have.
they’re a great use in surfacing information that is discussed and available, but might be buried with no SEO behind it to surface it
This is what I’ve seen many people claim. But it is a weak compliment for AI, and more of a criticism of the current web search engines. Why is that information unavailable to search engines, but is available to LLMs? If someone has put in the work to find and feed the quality content to LLMs, why couldn’t that same effort have been invested in Google Search?
If someone has put in the work to find and feed the quality content to LLMs, why couldn’t that same effort have been invested in Google Search?
I’d rather a world where 10 companies can compete with google search with AIs, than where they dump money into a monopoly.
If you don’t feel like discussing this and won’t do anything more than deliberately miss the point, you don’t have to reply to me at all.
The content is not unavailable to search engines. AI LLMs simply are better at surfacing it. I don’t know what point you were trying to make that I missed, it wasn’t on purpose, I assure you.
AI LLMs simply are better at surfacing it
Ok, but how exactly? Is there some magical emergent property of LLMs that guides them to filter out the garbage from the quality content?
Yeah. Money. Google has an incentive to make search results less accurate to get you to click around and interact with more ads. As it currently stands, AI models aren’t inserting advertisements; though I suspect that’s only a matter of time.
And that’s more or less what I was aiming for, so we’re back at square one. What you wrote is in line with my first comment:
it is a weak compliment for AI, and more of a criticism of the current web search engines
The point is that there isn’t something that makes AI inherently superior to ordinary search engines. (Personally I haven’t found AI to be superior at all, but that’s a different topic.) The difference in quality is mainly a consequence of some corporate fuckery to wring out more money from the investors and/or advertisers and/or users at the given moment. AI is good (according to you) just because search engines suck.
If you are just blatantly copying art, well yeah you’re stealing it.
ah yes it’s reactionary to checks notes not support the righteous biggest bubble since dotcom era
you okay out there bud?
You might want to look up the definition of reactionary. Because that’s…exactly what it means. To oppose reform/advancements.
You okay there bud?
In political science, a reactionary or a reactionist is a person who holds political views that favor a return to the status quo ante—the previous political state of society—which the person believes possessed positive characteristics that are absent from contemporary society.
Congratulations – Currently you and 18 others are not smarter than an average high schooler.
Opposing actual fraud isn’t what reactionary means.
You’ve got a pretty high bar of proof for proving “actual fraud”…
You can’t provably say that this is a “bubble” as claimed. The tools do what they purport to do. Where’s the fraud?
It’s not remotely within the realm of plausibility that Sam Altman genuinely believes any of the horseshit he spews. (And that’s ignoring that they gained their funding by lying about the core intent of their organization by pretending to be serving the public interest and not profiteering.)
It’s not remotely within the realm of plausibility that Sam Altman genuinely believes any of the horseshit he spews.
Welcome to earth. That’s basically every business ever, and you’ll quite literally never be able to prove that in court; which is the litmus test for this claim.
AI can give me a blueprint for my logic. Then I, as a developer, make the code run. Cuts my scripting time in half.
Rofl. As a developer of nearly 20 years, lol.
I used copilot until finally getting fed up last week and turning it off. It was a net negative to my productivity.
Sure, when you’re doing repetitive operations that are mostly copy paste and changing names, it’s pretty decent. It can save dozens of seconds, maybe even a minute or two. That’s great and a welcome assist, even if I have to correct minor things around 50% of the time.
But when an error slips through and I end up spending 20 minutes tracking down the problem later, all that saved time vanishes.
And then the other times where my IDE is frozen because the plugin is stuck in some loop and eating every last resource and I spend the next 20 minutes cursing and killing processes, manually looking for recent updates that hadn’t yet triggered update notifications, etc… well, now we’re in the red, AND I’m pissed off.
So no, AI is not some huge boon to developer productivity. Maybe it’s more useful to junior developers in the short term, but I have definitely dealt with more than a few problems that seem to derive from juniors taking AI answers and not understanding the details enough to catch the problems it introduced. And if juniors frequently rely on AI without gaining deep understanding, we’re going to have worse and worse engineers as a result.
I know tons of full stack developers who use AI to GREATLY speed up their workflow.
https://www.cio.com/article/3540579/devs-gaining-little-if-anything-from-ai-coding-assistants.html
How dare you bring sources into this opinion!
Let me guess. Dumped by an art girl and anxious about the $600 you invested?
How’s he wrong?
Did you actually listen to what he said or are you just reading the headline and making it fit another narrative to respond to?
Because he also said he thinks it’s going to change the world, he just hates the marketing BS that’s overhyping it.
Probably because, as anyone who’s actually used AI knows, it has some core weaknesses. But the marketers are happy to
gloss over thatlie and just say that it will be able to do nearly anything.He said it’s interesting, but to give it five years to see how it’s actually useful, which is probably the most sane take you can have about AI imo.
It will be interesting when the bubble pops, because that’s probably when we’ll see the useful things it is actually good at
But, it also means we get Sam Altman as the next Elon Musk if he cashes in before the pop. And whatever other tech bros do the same. More filthy-rich men with the emotional maturity of a 12 year old.
Which is how new technologies tend to go see what sticks after exploring what is possible. So it shouldn’t be surprising that ai is goong through the motions, but it is getting annoying how fast it is ruining functioning systems by being jammed in with no guardrails.
Summarizing documents, writing documents you don’t want to (within reason), and… whatever the hell Neuro-sama is doing on Vedal’s channel, are like the only ones i’ve found so far that kind of work. And I guess image generation.
It’s amazingly good at moderating user content to flag for moderator review. Existing text analysis completely falls down beyond keyword filtering tbh.
It’s really good at sentiment analysis. Which is great for things like user reviews. The Amazon ai notes on products are actually brilliant at summarizing the pros and cons of a product. I work for a holiday let company and we experimented with using it to find customers we need to follow up with and the results were amazing.
It smashes other automated translating services as well.
I use it a lot as a programmer to very quickly learn new topics. Also as an interactive docs that you can ask follow up questions to. I can pick up a new language as I go much faster than with traditional resources.
It’s honestly a complete game changer.
It’s honestly a complete game changer.
It is, both in good and bad ways. The problem, as Linus and others here are pointing out, is that marketing pushes the good and downplays/ignores the bad, so there’s going to be a rough adjustment period as people eventually see through the BS and find the issues, and the longer that takes, the harder things will crash.
There are plenty of good uses of modern AI approaches, they’re just far fewer than the ones being marketed these days.
The one place where I sincerely hope it takes root and succeeds is in medicine. Having better drugs, helping to identify potential problems or diseases, identifying health patterns (all with human review and proper trials, naturally)…
It’s not even close to the magical AGI that tech bros are promising, but it is good at digesting data, and science and medicine are full of that. Plus, given how overworked doctors and nurses can be, having a preliminary analysis from a computer that doesn’t get tired or overworked seems like it would probably help with accurate diagnosis.
Speaking as someone who worked on AI, and is a fervent (local) AI enthusiast… it’s 90% marketing and hype, at least.
These things are tools, they spit out tons of garbage, they basically can’t be used for anything where the output could likely be confidently wrong, and the way they’re trained is still morally dubious at best. And the corporate API business model of “stifle innovation so we can hold our monopoly then squeeze users” is hellish.
As you pointed out, generative AI is a fantastic tool, but it is a TOOL, that needs some massive changes and improvements, wrapped up in hype that gives it a bad name… I drank some of the kool-aid too when llama 1 came out, but you have to look at the market and see how much fud and nonsense is flying around.
As another (local) AI enthusiast I think the point where AI goes from “great” to “just hype” is when it’s expected to generate the correct response, image, etc on the first try.
For example, telling an AI to generate a dozen images from a prompt then picking a good one or re-working the prompt a few times to get what you want. That works fantastically well 90% of the time (assuming you’re generating something it has been trained on).
Expecting AI to respond with the correct answer when given a query > 50% of the time or expecting it not to get it dangerously wing? Hype. 100% hype.
It’ll be a number of years before AI is trustworthy enough not to hallucinate bullshit or generate the exact image you want on the first try.
Half of the people here linus included must have never use stable diffusion
He isnt wrong. This comes from somebody who technically uses ai daily to help develop ( github copilot in visual studio to assist in code prediction based on the code base of the solution ), but AI is marketed even worse than blockchain back in 2017. Its everywhere, in every product, even if it doesnt have ai or has nothing to do with it. Monitor ai shit? Mouse with ai? Hell, ive seen a sketch of a fucking toaster with ‘ai’.
There is shit like microsoft recall, apple intelligence, bing co pilot, office co pilot, …
All of those are just… Nothing special or useful. There are also chatbots which bring nothing new to the table either.
Everyone and everything wants to market there stuff with ai and its disgusting.
Does that mean that current ai tech cant bring anything to the table? No, it totally can, but 90% of ai stuff out there is, just like linus says, marketing bullshit.
AI is nothing more than a way for big businesses to automate more work and fire more people.
and do that at the expense of 30+ years of power reduction and efficiency gains, to the point that private companies are literally buying/building/restarting old power plants just to cover the insane power demand, because literally operating a power plant is cheaper than paying the energy costs.
For the common every day person its 3d tv and every other bullshit fad that burned brilliantly for all of 3 seconds before snuffing itself out, leaving people to have had paid for overpriced garbage thats no longer useful.
AI is nothing more than a way for big businesses to automate more work and fire more people.
All technology in human history has done that. What are you proposing? Reject technology to keep people employed on inefficient tasks?
At some point people need to start thinking that is better to end capitalism that to return to monke.
game devs gonna have to use different language to describe what used to be simply called “enemy AI” where exactly zero machine learning is involved
Logic and Path-finding?
I had a professor in college that said when an AI problem is solved, it is no longer AI.
Computers do all sorts of things today that 30 years ago were the stuff of science fiction. Back then many of those things were considered to be in the realm of AI. Now they’re just tools we use without thinking about them.
I’m sitting here using gesture typing on my phone to enter these words. The computer is analyzing my motions and predicting what words I want to type based on a statistical likelihood of what comes next from the group of possible words that my gesture could be. This would have been the realm of AI once, but now it’s just the keyboard app on my phone.
There’s a name for it the phenomenon: the AI effect.
The approach of LLMs without some sort of symbolic reasoning layer aren’t actually able to hold a model of what their context is and their relationships. They predict the next token, but fall apart when you change the numbers in a problem or add some negation to the prompt.
Awesome for protein research, summarization, speech recognition, speech generation, deep fakes, spam creation, RAG document summary, brainstorming, content classification, etc. I don’t even think we’ve found all the patterns they’d be great at predicting.
There are tons of great uses, but just throwing more data, memory, compute, and power at transformers is likely to hit a wall without new models. All the AGI hype is a bit overblown. That’s not from me that’s Noam Chomsky https://youtu.be/axuGfh4UR9Q?t=9271.
I’ve often thought LLMs could replace all of the C-suites and upper and middle management.
Funny how no companies push that as a possibility.
I almost expect that we’ll see some company reveal it has been letting an AI control the top level decision making for the business itself, including if and when to reveal the AI.
But the funny thing will be that all the executives and board members still have jobs and huge stock awards. They will all pat each other on the back for getting paid more money to do less work, by being bold and taking a risk to let the computer do half their job for them.
I dunno about him; but genuinely I’m excited about AI. Blows my mind each passing day ;)
I work at a company big into AI. We build our own models. Our senior management drank the Kool-Aid. We don’t have search on our Intranet any more, just LLM chatbots.
Our TLS certificate expired last week on our main web page. I tried to find the contact details for the team responsible and the thing just hallucinated e-mail addresses.
Needless to say, I’m less excited than you.
This is hilarious
Holy fuck my sides hurt laughing. Sending a virtual IT bro hug.
That sounds insane
“duh.”
No AI is a very real thing… just not LLMs, those are pure marketing
The latest llms get a perfect score on the south Korean SAT and can pass the bar. More than pure marketing if you ask me. That does not mean 90% of business that claim ai are nothing more than marketing or the business that are pretty much just a front end for GPT APIs. llms like claud even check their work for hallucinations. Even if we limited all ai to llms they would still be groundbreaking.
Korean SAT are highly standardized in multiple choice form and there is an immense library of past exams that both test takers and examiners use. I would be more impressed if the LLMs could show also step by step problem work out…
Claud 3.5 and o1 might be able to do that; if not, they are close to being able to do that. Still better than 99.99% of earthly humans
You seem to be in the camp of believing the hype. See this write up of an apple paper detailing how adding simple statements that should not impact the answer to the question severely disrupts many of the top model’s abilities.
In Bloom’s taxonomy of the 6 stages of higher level thinking I would say they enter the second stage of ‘understanding’ only in a small number of contexts, but we give them so much credit because as a society our supposed intelligence tests for people have always been more like memory tests.
and that 10% isnt really real, just a gabbier dr.sbaitso
Idk man, my doctors seem pretty fucking impressed with AI’s capabilities to make diagnoses by analyzing images like MRI’s.
then you are a fortunate rarity. most postd about the tech complain about ai just rearranging what it is told and regurgitating it with added spice
I think that is because most people are only aware of its use as what are, effectively, chat bots. Which, while the most widely used application, is one of its least useful. Medical image analysis is one of the big places it is making strides in. I am told, by a friend in aerospace, that it is showing massive potential for a variety of engineering uses. His firm has been working on using it to design, or modify, things like hulls, air frames, etc. Industrial uses, such as these, are showing a lot of promise, it seems.
thats good. be nice if all the current ai developers would aim that way
Seems generous, might be more like 5% reality.