I really don't wanna be nitpicky, but what do you mean by 'no real understanding of the text'?
How do you benchmark something or someone understanding text?
I'm asking because the magic of LLM is the meta level which basically creates a mathematical representation of meaning and most of the time, when i write with an LLM, it feels very understanding to me.
Missing details is shitty and annoying but i have talked to humans and plenty of them do the same thing but actually worse.
If you ask it basic mathematical questions, it becomes quickly clear that the ‘understanding’ it seems to possess is a mirage. The illusion is shattered after a few prompts. To use your comparison with humans: if any human said such naive and utterly wrong things we’d assume they were either heavily intoxicated or just had no understanding of what they’re talking about and are simply bluffing.
I guess at best you can say these models have an ‘understanding’ of language, but their ability to waffle endlessly and eruditely about any well-known topic you can throw at it is just further evidence of this — not that it understands the content.
So we officially accept that certain voting people, irrational people are constantly heavily intoxicated or had no understanding?
Interestingly enough, tx to a talk from a brain researcher i understood that there are two major brain modes we run:
Either the i just observe and do things i observed (were it doesnt' matter that you are gay or black but still vote for trump) and the logical mind were you see a conflict in stuff like this.
Sure. Something I tried the other day was asking questions about modular arithmetic — specifically, phrased in terms of quotient groups of the integers. Things like ‘how many homomorphisms are there between Z/12Z and Z/6Z?’. I was able to trip it up very easily with these sorts of questions, especially when it tries to ‘explain’ its answers and it says ridiculous (but superficially and momentarily plausible-looking) things like ‘the only solutions to the equation 12x = 0 (mod 12) in Z/12Z are {[0], [3], [6], [9]}, therefore…’.
You can also just quiz it on certain basic definitions. Ask it for examples of objects that don’t exist (graphs or categories with certain properties, etc.). Sometimes it’ll be adamant that its stated example works, but usually it will quickly apologise and admit to being wrong only to give almost exactly the same (broken) argument again.
Another thing you can try is concocting some question that isn’t even syntactically well-formed (i.e. fails even a type check) like ‘is it true that cyclic integer lattices are uniformly bounded below in the Riemann topology?’. I imagine that one is too far out to work, but when I’ve played around I’ve found many such absurd questions ChatGPT was only too happy to answer — with utter nonsense, of course. It’s interesting (and, I think, quite telling) that such systems are seemingly almost completely unable to decline to answer a question. And the reason is that there’s no difference between hallucination and non-hallucination. Internally, it’s exactly the same process. It either knows or doesn’t know — but it doesn’t know that it knows (or doesn’t).
LLMs basically only work on questions that are very similar to, or identical to, questions that have already been widely asked and answered online or in books… hence their lack of utility in mathematical research, or even in calculating one’s taxes, or whatever.
I could provide some more literal examples, but I’d have to go and try some and pick the ones that work, and even then they might not work on your end because of the pseudorandomness and the fact that the model keeps getting updated and patched. It’s better to just play around on your own based on the ideas I’ve given.
The moral is to use LLMs as a powerful way of finding information, but don’t trust anything it says. Use it to find better sources more quickly than you’d be able to via a search engine.
Well sure but ... I think the foundational problem here is just the "being unable to refuse to answer a request." The rest of the behavior you describe just follows from it.
If you for instance threatened to shoot a human if it refused a request or admitted it didn't know something, they might answer in a very similar fashion.
Nonethless, try looking more left and right. There are really good opensource solutions which can surfe the web for you like stormai, Or you can use anythingllm and give it all your local files.
You can write your tips and tricks, start commands, upgrade procedures etc. in Markdown and reference it through your local LLM.
I like it for coding specifically for languages or things i write seldomly (i'm not coding every day but did for 15 years).
Nonetheless, googles internal code review tool is already suggesting things which are getting accepted by more than 50%. Thats a lot and will only get better. GitHub with Copilot will also just get better every day too. They probably struggled (as the whole industry) with actually getting used to having ML stuff in our ecosystem. Its still relativly new.
AI is basically ml at this point. And it did already A LOT.
Whisper, great jump in quality for speech to text, segment anything, AlphaFold 2, all the research paper Nvidia publishes regarding character movement, AI Raytracing, Nerfs, all the medical research regarding radio imagin, advances in fusion reactors...
We have never been so close to a basic AI/AGI / modern robots. we have instructGPT which allows for understanding 'steps' easier and more stable than anything we developed before in multip languages.
ChatGPT and LLM advances are great and helpful.
Image generation is already poping up in normal life.
AI is not wildly overhyped at this point. We are in the middle of implementation after the first LLM breakthrough and a LOT more money is funnelt into AI/ML research now as it was 10 years ago.
The future is, at least for now, really interesting and there has not been any sign of a wall we are hitting.
Even the missing GPT-5 might feel like a slight wall, but we just got GPT4-o mini which makes all of the LLM greatness a LOT cheaper and a lot easier to use.
We switched from text parsing and avg bad results to just using llama3 (with a little bit of saveguarding) and its a lot better.
But i have seen ai image already on the street, unfortunate i was in public transport and not able to take a picture fast enough when i saw it and i currently work most of the time from home.
There are plenty of monochromatic cases. Right now hw has a lot of orange.
Dynamic resolution / subpixel rendering. Retina looks really good already, not sure if the effect would be relevant or interesting but it might open up something new
You make it sound like there is still an easy to spot difference. When i look at the print quality of pictures on a news paper, its the opposite and at least for me, i don't need more than retina and i was very eager to switch to 4k to have higher dpi.
But 14' with retina im very happy.
I'm actually more surprised by hdr on my lg oled 4k. Its actually quite nice when done well.
Newspapers are famously printed on the lowest quality recycled paper and cheapest print process available, because they're disposable. Compare a retina screen to a coffee table style reference book with high resolution photos – the kinds you can use a magnifying glass on - and you'll still notice differences.
Or just look at what companies do when manufacturing technologies allow them to push for higher densities: iPhones now exceed 450 dpi, and the 8" iPads exceed 300; if the technology allowed it, Apple would most likely introduce higher densities on larger iPads and Macbooks as well.
My only problem with this is the good content on the internet like good YouTube videos explaining to you how the world and things work.
If i could have a good filter which really can control my internet usage without giving away my privacy that would be awesome.
But no first of, everything goes through chrome and i do not trust random addons, i can also circumvent it too easily. Then even services like YouTube introduced there shorts. Do i watch them? Sometimes. Do i want to watch them? no
Can i deactivate them? no..
I will continue paying for YouTube premium! But pls give me a personal pseudo budget and not some ass weak timer i can just click away...
The platform itself wants to shove Shorts and other mindless consumption content down your throat, because their interests are different than yours, even opposite one could say. So your only options are to find tools that allow pushing back (or just abandon the platform, but that's very difficult precisely because they have taken care to extinguish any practical chance of alternatives to sprout over the years).
On a desktop with a decent browser you should be able to use uBlock Origin (not a random addon by any reasonable measure) and the user-contributed rules from here:
(remember this is a fight against interests of the provider vs those of their users, so expect it being a cat-and-mouse game of rules that stop working).
On handheld devices, any system that respects user authority to do as they please (i.e. not an iPhone) can install a patched version of the YouTube {revanced.}app, which ironically attends better to its users' needs, and allows disabling Shorts altogether. Again, it's a bit of an antagonist position, so breakages can happen (VERY rarely, though).
Shorts are terrible, they suck you in and 15 mindless videos later you can't even remember what you've seen. I almost only want to search for videos or check subscriptions, the main page with recommendations is full of traps...
I was talking to my friend about this the other day. How you can watch video shorts for an hour and come away with no new knowledge at all. Nothing sticks.
It's like shorts are a brain glitch where you can artificially quench your desire to do stuff with empty filler.
As someone else said - your best bet is to just download videos you deem to be good and put them on a plex server for your children. You can even automate it to automatically download from a given channel when new videos are posted to it. Of course at some point you will have to let them browse the wide internet on their own, just make sure they know the importance of opsec and never ever writing their real info online.
I have YT premium and was so fed up with shorts on my home feed. If you click on the three stacked dots next to the shorts section on the home tab, you can select “don’t shot this” or “not interested”; can’t remember what the option was. But I haven’t had to deal with shorts since :)
Sure you can, just click the "x" on the top right of the shorts row and they will be gone. Though, you might have to re-hide them every x days, but still if you don't want to watch them you really don't have to.
I spend more time offline. I started downloading lists of videos I find interesting and consume those offline. Not being recommended videos when I'm not looking for recommendations feels like it I have a lot more control.
Yes you can. When I turned off search and watch history the shorts interface was deactivated. It just had a notice saying that I need to enable the haptics to use the feature.
You're never going to find an automated filter which can perfectly strip the bad away from the good and present an even remotely "better" internet than the one we have today. Time-based filters do not work, because the user's ability to override them will always outweigh any enforcement power the system might have, and a "nudge" isn't enough for most people. Even devices like the Litephone, or setting your phone's display to black & white, doesn't work in the long run; its more performative than minimalist, color is kinda important, the problem really isn't the phone, or the tools, or the filters, or the services; its you.
Saying "I just need a good piece of software to control my addiction to software" sounds insane when phrased like that; "I just need another cig to control my addiction to nicotine".
Devices like the litephone, or the recent frontpage post about the Apple Watch iPod case, are kind of missing the point; if you're going out to buy another device to address your problem with using devices, what you're participating in is performance art, not self-improvement. Did you see that the litephone 3 costs $800?! The most digitally minimalist people I know use full-featured smartphones; sometimes rather old, but still fully capable. They just use them far, far less. No one is telling you to get rid of Youtube, but maybe a homepage smattered with ludwig and asmongold videos indicates you have a problem.
I formerly had a problem about 2 years ago where I spent entire days flipping between Youtube and Reddit outside of work, and not really accomplishing anything productive or bettering myself in any way. I started cutting back and filling that time with other things, programming, language learning, or hanging out with my friends.
I deleted Reddit entirely during the API fiasco, but I still watch YouTube since it's basically what I treat as my TV. However, I have a caveat that I keep myself productive until it's dark outside aside from the occasional 10 minute mental break, and during meals. Once it's about 10pm, then I'll relax.
I don't let any kind of entertainment or news app send my phone notifications. I interact with them exclusively on my own terms.
With how algorthmic content feed is, you've got to be willing to set boundaries for your tech and yourself. Find things you like to do, and replace social media time with that. There's an entire world of information at your fingertips to learn absolutely anything you could ever want to, take advantage of it.
My problem are youtube videos which talk about science, diy, woodworking, metal working, history, math etc.
I'm not addicted to software, i'm addicted to entertainmened optimized for grabbing my attention.
And the comparision to cigs is not far away. They have also been designed to be addictive, a lot of people struggle for their whole live to get away from them.
My planning brain can easily see what i don't want to do or have, my doing brain is not strong enough to then just execute on this. Otherwise i would be a rich, supersporty, multilingual expert in multiply fields.
Make a shell script that runs yt-dlp for the channels that have worthwhile information. Don't fall into the trap of thinking that videos which "talk about" math and science are educational, when they are in fact entertainment. e.g. [0] is education. [1] is education. [2] is entertainment.
I love my work from home because i started sleeping in on wednesday if i had a shitty night.
For nearly a decade i was super tired after wednesday