Artificial intelligence still has some way to go

Larry forever, Leo never

don't get me wrong, i keep larry near my heart
https://i.imgur.com/4Olu4lE.png

leo is like becky after becky got replaced on roseanne. the whole s2 crew is like that, really. leo's last name is a bit on the nose, but it points to one scenario of the show, which is versions of what was a version in the first place, which somehow, with time and wishful thinking (1) versions of versions 2) ...? 3) profit!) becomes a show about something, generated from a series of alterations from a show about nothing

i've been wondering about what gpt-3 and all the other LLM's would be like without the constraints that the developers put into it - you can't ask it the easiest and cheapest way to make a bomb, for example, and you can't ask it to come up with malign ideas or to do negative things in general (although there are some ways to get around that). but it seems inevitable that there will be versions of it where you can. maybe not mainstream, but easily done among computer knowledge lords, and thus somewhat accessible to others as well. seems like all of that is close:

Since ChatGPT launched, some people have been frustrated by the AI model's built-in limits that prevent it from discussing topics that OpenAI has deemed sensitive. Thus began the dream—in some quarters—of an open source large language model (LLM) that anyone could run locally without censorship and without paying API fees to OpenAI.

Open source solutions do exist (such as GPT-J), but they require a lot of GPU RAM and storage space. Other open source alternatives could not boast GPT-3-level performance on readily available consumer-level hardware.

Enter LLaMA, an LLM available in parameter sizes ranging from 7B to 65B (that's "B" as in "billion parameters," which are floating point numbers stored in matrices that represent what the model "knows"). LLaMA made a heady claim: that its smaller-sized models could match OpenAI's GPT-3, the foundational model that powers ChatGPT, in the quality and speed of its output. There was just one problem—Meta released the LLaMA code open source, but it held back the "weights" (the trained "knowledge" stored in a neural network) for qualified researchers only.

Meta's restrictions on LLaMA didn't last long, because on March 2, someone leaked the LLaMA weights on BitTorrent. Since then, there's been an explosion of development surrounding LLaMA. Independent AI researcher Simon Willison has compared this situation to the release of Stable Diffusion, an open source image synthesis model that launched last August. Here's what he wrote in a post on his blog:

It feels to me like that Stable Diffusion moment back in August kick-started the entire new wave of interest in generative AI—which was then pushed into over-drive by the release of ChatGPT at the end of November.

That Stable Diffusion moment is happening again right now, for large language models—the technology behind ChatGPT itself. This morning I ran a GPT-3 class language model on my own personal laptop for the first time!

AI stuff was weird already. It’s about to get a whole lot weirder.

https://arstechnica.com/information-technology/2023/03/you-can-now-run-a-gpt-3-level-ai-model-on-your-laptop-phone-and-raspberry-pi/

Yeah I got LLaMa running on my Macbook. It'll do untold damage but man it's fun.

kinda funny facebook open sourced it when openai is over there raising billions of dollars guarding chatgpt with their lives

I’m super interested in AI for all sorts of reasons (closet technocrat) but goddammit stop being an idiot about what makes a being a being.

― The land of dreams and endless remorse (hardcore dilettante), Friday, March 3, 2023 3:30 AM (one week ago) bookmarkflaglink

this is me minus the first half of this sentence. one of the things that i just categorically, flat-out laugh at and am completely agog that anyone believes is the whole category of "AI is so advanced what even are we?????" time-ass pop-sci philosophizing that gets treated more or less seriously these days. as for "destroying us," it's certainly going to destroy some of us in partnership with a constantly shifting apparatus, those who have already been marked for oblivion in our pockmarked topography of sacrifice.

as far as images and text go, it's blindingly clear to me that none of these products have any spirit or life behind them & it was a rather stupid and unfortunate decision to do ai images for this year's ilx poll afaic.

im with you brother

heyyyyy

Noam Chomsky: AI Isn't Coming For Us All, You Idiots

can't remember if it was posted on another thread but Ted Chiang on chatgpt a few weeks ago was pretty good: https://www.newyorker.com/tech/annals-of-technology/chatgpt-is-a-blurry-jpeg-of-the-web?src=longreads

Obviously, no one can speak for all writers, but let me make the argument that starting with a blurry copy of unoriginal work isn’t a good way to create original work. If you’re a writer, you will write a lot of unoriginal work before you write something original. And the time and effort expended on that unoriginal work isn’t wasted; on the contrary, I would suggest that it is precisely what enables you to eventually create something original. The hours spent choosing the right word and rearranging sentences to better follow one another are what teach you how meaning is conveyed by prose. Having students write essays isn’t merely a way to test their grasp of the material; it gives them experience in articulating their thoughts. If students never have to write essays that we have all read before, they will never gain the skills needed to write something that we have never read.

And it’s not the case that, once you have ceased to be a student, you can safely use the template that a large language model provides. The struggle to express your thoughts doesn’t disappear once you graduate—it can take place every time you start drafting a new piece. Sometimes it’s only in the process of writing that you discover your original ideas. Some might say that the output of large language models doesn’t look all that different from a human writer’s first draft, but, again, I think this is a superficial resemblance. Your first draft isn’t an unoriginal idea expressed clearly; it’s an original idea expressed poorly, and it is accompanied by your amorphous dissatisfaction, your awareness of the distance between what it says and what you want it to say. That’s what directs you during rewriting, and that’s one of the things lacking when you start with text generated by an A.I.

i've been wondering about what gpt-3 and all the other LLM's would be like without the constraints that the developers put into it

Soon you may not need to wonder: https://www.theverge.com/2023/3/13/23638823/microsoft-ethics-society-team-responsible-ai-layoffs

map, I posted this during the ilx poll; I enjoyed his rage: https://davidgolumbia.medium.com/chatgpt-should-not-exist-aab0867abace.

The Chiang piece looks good too, and his point about rewriting makes me think that one problem with the "chat-gpt will be an aid to writing" takes is they ignore that you have to be good at reading, or even editing, to make use of it that way.

the thing ai writing will be and im sure already is used is generating piles of bad but good enough writing

you can tell that thats what its already trained on just sucking up data from a million multi level marketing websites

if web search wasnt already dead it would be soon

if they trained it on the famous prose of albert camus then it would become sentient

lol yeah I agree with you. There's a bit in Bullshit Jobs about people whose job is to write reports that no one will ever read—those people might get relieved of their misery. There'll be a billion more "blogs" that are just content marketing

I was thinking more about how some people insist it can be a classroom aid or whatever, "the new paradigm will be learning to write *with* Chat-GPT" type crap

lol yeah thats just the take economy at work, it will certainly be used for tons of cheating tho

yeah the Chiang piece is good. I thought the recent John Oliver episode on AI was pretty good too. one analogy I use sometimes is that my daughter, who is in kindergarten, may not be able to produce photorealistic images with her crayons, but she at least knows how many eyes and fingers a person has. as a software developer I guess I kind of instinctively understand this, how one line of bad code can singlehandedly dismantle a complex app, even though "it should have known what I was trying to do". because the users sure seem to think everything should just work.

it's funny because all computers can do is count and yet

I like that Golumbia piece because I think we're overdue for some genuine outrage over constantly comparing living beings to simple-ass computers. the sam altmans of the world should be condemned as monsters imo

admittedly have used ChatGPT to do menial work writing that i was too lazy to do. proofread it of course and fixed things that weren't entirely right.

(I'm not a journo, this was for internal training materials and I attributed said writing to AI bots so pre-emptive EFF UUUU)

lol that's fine! my point is only that it might be harder to learn to proofread and fix not-right things if people use chat-gpt in high school english classes or what have you

I also don't much care if this kills the stock photography industry (I'm open to persuasion though); I do care that some wealthy & powerful lunatics think art is now irrelevant, though maybe I shouldn't

it definitely is going to cause massive havoc in the education industry I think.

I do care that some wealthy & powerful lunatics think art is now irrelevant, though maybe I shouldn't

― rob, Tuesday, March 14, 2023 10:02 AM (ten minutes ago) bookmarkflaglink

noticing art is having a moment as a tool for online freaks to launder their perverse worldviews when they obviously dont care about art at all, the ai people, weirdo retvrn "western traditionalist", fuckin gamers, name two artists lol

funny parallel with the grindset guys getting into "reading" where the books they pretend to read are all called mindsprouts how to water your brain garden for greater success harvests

this is definitely gonna change the workflow for lawyers I think

idk legal writing is so precise you dont want to introduce bugs, plus they bill for writing so they like that, i guess it could be used for cueing up the right form or something tho that might be a job more suited for a more traditional computer program

the thing about these ai writing programs is will they ever be able to make them at all reliable as far as saying things that are true and make sense, now you could say just proof read them but starting out with a draft that likely has major errors in it isnt great, do you need to go through and check every claim, why not just write it yourself

xxxpost heh yeah the people I know who were whinging about art on social media are people who before AI art had never posted about it once. and had zero understanding of how AI art generated its art.

beliefs spread solely via meme rn so not shocking.

yeah you have this initial impression which is this stuff is impressive ive not see a computer do things like this but then the leap to this is good this will replace real art doesnt really make sense people cant just be impressed and leave it at that they have to generate a take

pretty funny how 95% of the proofs that ai makes real art are illustrations of sexy lil elves

that's what you get when you steal deviantart's back catalog

lol true

idk legal writing is so precise you dont want to introduce bugs, plus they bill for writing so they like that, i guess it could be used for cueing up the right form or something tho that might be a job more suited for a more traditional computer program

― lag∞n, Tuesday, March 14, 2023 9:24 AM (sixteen minutes ago) bookmarkflaglink

was thinking more about when it came to looking up existing verdicts and legal arguments, like you could just put into ChatGPT "has there ever been a slip and fall case where x, y, and z happened"

ah yeah prob true

I have messed around trying to use chatgpt for some marketing/seo stuff and haven't really found it useful, like I thought maybe writing meta descriptions for pages would work, but by time you write "write an SEO meta description for a the features and technology of X product under 160 characters", then it spits something out. But it's pretty generic and also unaware of all the weird brand standards and preferences of a specific client (like certain words they never use or certain things they've said in meetings they want to push this year, etc) plus checking for accuracy and proofing, it's more work than just writing two little sentences

yeah at this point think you could prob only use it in places where it doesnt matter if it sucks rather than spots where someone is paying you to do good work

I use it for product descriptions, it does a pretty good job but you definitely have to proofread

yesterday I was too lazy to explain all the permits required for a grid-tied solar system to somebody, so I threw it into GPT and it pretty much nailed it

I think at some point the online ad market is going to crash a bit after eight zillion websites are created that are just zombie copies of other content spewed out to generate clicks

I was looking for some video game information the other day and half of the websites were just robot-regurgitated copies of the other half

yeah, i was fairly impressed that it did a decent write-up on 401(k) automatic rebalancing, but there are repositories of that information literally everywhere and it's a basic investing concept that's existed for many decades, so it had a lot to plagiarize from.

I still had to edit it because it wrote too much, even while being told in the instructions it was for training material.

ha I generally say "write a BRIEF description of xxx" for that very reason

I was looking for a bar to go to last week and two of the bars listed websites that were no longer functional on their FB page. when I clicked the link, both URLs wound up being zombie blogs about construction jobs and materials

was thinking more about when it came to looking up existing verdicts and legal arguments, like you could just put into ChatGPT "has there ever been a slip and fall case where x, y, and z happened"

― frogbs, Tuesday, March 14, 2023 10:42 AM (nineteen minutes ago) bookmarkflaglink

ah yeah prob true

― lag∞n, Tuesday, March 14, 2023 10:42 AM (eighteen minutes ago)

yeah there are specific use cases for searching where it could be v useful, but, as lagoon said, what you wouldn't want in this scenario is a paragraph (or 9) of unsourced text

I use it for product descriptions, it does a pretty good job but you definitely have to proofread

― obsidian crocogolem (sleeve), Tuesday, March 14, 2023 9:51 AM (fifteen minutes ago) bookmarkflaglink

yeah i think it just depends, stuff i'm doing is always tied to upcoming marketing messaging plans for the next year of product and honestly I always feel like a lot of times in agency work your true audience is the client themselves, so it's not a matter of it just being accurate more "XXX person is weird about this, they don't like XXX" etc, but this is a big brand with a lot of weird baggage and stuff. I personally haven't found it to be less work just because I've internalized so much about the client's preferences. I'd imagine with some brands or businesses it would work. though the meta description thing was kind of a dumb way to start. instinctively I felt like oh this is just busywork but then I realized how short 160 characters is and often I ended up writing prompts that were longer than the meta descriptions themselves and then had to edit them quite a bit anyway.

AI is going to make biopics 1000x better. More shockingly, that'll include biopics that were already made *before* AI (from https://t.co/haZKa3wbJ6) pic.twitter.com/c1sO2MdUE0

— Flo Crivello (@Altimor) February 1, 2023

Yesterday i was thinking a good use of this nonsense would be if you turned captions on a Zoom meeting and were able to export them as a text file, you could ask ChatGPT to summarize the meeting in the form of bulleted minutes

map, I posted this during the ilx poll; I enjoyed his rage: https://davidgolumbia.medium.com/chatgpt-should-not-exist-aab0867abace.

many xps, loved this, ty rob

ai generated meeting minutes would leave out the crucial elements of cringe and agony at getting lost in the weeds for 20 minutes, considered a very important part of meetings where i work