Artificial intelligence still has some way to go

clearly we have learned nothing from the tragic malpractice of DR_SBAITSO.exe

the irony, i guess, is that ELIZA was one of the first chatbots, in the 60's, and was designed to emulate Rogerian therapy

xp i didn't know about Dr. Sbaitso!

i guess the Rogerian kind of "and how does that make you feel?" kind of therapy is a natural fit for chatbots with limited capabilities

yes!! we had this on our computer when I was growing up! god my brothers and I would spend hours typing dirty words into that thing

https://i.imgur.com/VfgMMzt.png

also, yet more proof that although the internet gets worse every single year, at least this means that the further back in time you go, the better it gets
https://archive.ph/20130111132657/http://www.x-entertainment.com/articles/0952/

because it's the future, this already exists
https://bert.org/2023/01/06/chatgpt-in-dr-sbaitso/

I think ai is actually a good fit for cognitive therapy because it is logical and analytic - but can it provide human empathy?

People fake empathy all the time. Why not a bot?

faking empathy is all a bot CAN do. it's the recipient that's objecting.

I guess sometimes empathy backfires too like
"I'm so tired after that walk in the prairie"

"really? I'm not tired at all! You must be an increasingly inferior individual to tire so easily!"

Also some humans give the worst &*^%&^% advice

emptythy

https://stablediffusionlitigation.com/

https://www.vice.com/en/article/xgyjm4/ai-writing-tools-like-chatgpt-are-the-future-of-learning-and-no-its-not-cheating

“I think it’s an increase in human capability moment that we’re looking at right now,” co-director at Deakin University’s Centre for Research in Assessment and Digital Learning, Phillip Dawson, told VICE.

“I think a student that graduates in five years’ time is going to be able to do so much more than what we are capable to do now because they’ll be using these sorts of tools.”

Dawson described ChatGPT as a writing tool and compared students using it to help them write essays to a pilot learning how to fly a modern plane.

“Yeah, you need to be able to use all the instruments and you need to know how all those work, but you also need to be able to do it when all those instruments fail. You still need to be able to land that plane.”

i think this is right.

i distinctly remember an early 2000s drunken front porch discussion i had with a friend, back in the days when i had friends and we had discussions. he said that in "the future", it wouldn't be about knowing things, it would be about knowing the tools that you could use to find out things. fluency with finding information would be more useful than just knowing things. i took a giant bong rip. yes. yes, man. he was right, and that future is already the past.

i think this ai stuff is fucking terrible. but it's not going away. the conversation will soon shift to how to make it better and to fix the things that are bad about it. but it's not going away. i can't change that, you can't change that. it's a blow to the ego, and maybe that, in itself, is a good thing

good thoughts

I read a very click baitty piece on AI and the end of writing but all the examples were basically "look at how AI apes this style of writing from the 1400s" and then went "this will write Ulysses with a prompt" which misses the point that Ulysses has been written and this would be pointless.

You will still need to memorise and store a lot of information in your head. There is a base of canon like knowledge -- whether that's a canon of poetry or scientific knowledge -- that we all need to work at.

AI will calculate and be an aid to coming out with interesting outputs, that as prompts we can use. We can may get to a good destination faster.

I remember that argument being said about Chess and Go. Then AI taught itself from first principles (the rules) with no pre-existing body of knowledge.

But it hasn't destroyed the enthusiasm for chess playing and competition.

Far from just providing useful outputs, in these areas AI has overtaken us in reaching a good destination (superhuman playing strength) and left us in the slow lane.

It's a good example really. In the end though there are tournaments, a public that is interested and watches and plays with the knowledge that a machine would beat the best human.

Funnily enough the outrage and alienation come from when we think a human is using AI under the covers to cheat another human being.

I finally gave in and had a go on ChatGPT. It didn't seem confident in offering an opinion on whether my child's grandmother (whom it later referred to as 'your grandmother') would like Super Mario Bros, and I had to correct its punctuation after some terrible run-on sentences. It then apologised and said it should've used semicolons, but when I asked which sentence I had been referring to it quoted the wrong one.
IDIOT!
When I told it which sentence I meant though, it did correct it properly. Hmm.

I think it's great that ChatGPT has exposed Nick Cave as totally hollow talent after someone asking it something like "write some shite knockoff Flann O'Brien lyrics about murdering women in the style of a sad old aging goth"

Certainly a bit of a moment of reckoning for secondary school teachers and university professors who assign essays that are basically "regurgitate this week's syllabus in your own words please". They are going to have to get more creative in their assignments and that's got to be a good thing!

who would have known, CNET

When internet sleuths discovered last week that CNET had quietly published dozens of feature articles generated entirely by artificial intelligence, the popular tech site acknowledged that it was true — but described the move as a mere experiment.

Now, though, in a scenario familiar to any sci-fi fan, the experiment seems to have run amok: The bots have betrayed the humans.

Specifically, it turns out the bots are no better at journalism — and perhaps a bit worse — than their would-be human masters.

On Tuesday, CNET began appending lengthy correction notices to some of its AI-generated articles after Futurism, another tech site, called out the stories for containing some “very dumb errors.”

An automated article about compound interest, for example, incorrectly said a $10,000 deposit bearing 3 percent interest would earn $10,300 after the first year. Nope. Such a deposit would actually earn just $300.

More broadly, CNET and sister publication Bankrate, which has also published bot-written stories, have now disclosed qualms about the accuracy of the dozens of automated articles they’ve published since November.

New notices appended to several other pieces of AI-generated work state that “we are currently reviewing this story for accuracy,” and that “if we find errors, we will update and issue corrections.”

which is cheaper - a writer and an editor a decent wage, or hire just an editor (probably not at a decent wage)

New notices appended to several other pieces of AI-generated work state that “we are currently reviewing this story for accuracy,” and that “if we find errors, we will update and issue corrections.”

stuff a bot would say

i want to say something positive. because, reading this thread, i think i give the wrong impression. i feel that way about a lot of stuff, a lot of topics.

i think there's ai and creativity don't have to be enemies. i don't think i'll ever be a person that uses it as part of the process, but i already think about it and think about it as something akin to google image search, back when google image search was good, something you can modify and play with and use in unexpected ways. a couple weeks ago i prompted one of the crappy free ones, craiyon, with "pink floyd album covers with babes and butts", just wanting to see what it would do.

https://i.imgur.com/8ePC0l0.png

i'm not sure what that is. and if and when i ever made a drawing or a painting from that, i would change a lot. there's no point to trying to replicate it. but the left butt is incredible, as it the bikini in the middle, and i love how the black swimwear on the right doesn't make sense. it's like a miniskirt on the beach, with grumpy neil young nearby. it's a good prompt for making art. it doesn't make sense to look at it (to me) and disregard it out of hand because the source was a computer. those horizontal lines on the left and his rectangle arm. that's weird, and good, and thought provoking to me

i love how all three of the figures on the right are taller than the black-clad figure on the left

which is cheaper - a writer and an editor a decent wage, or hire just an editor (probably not at a decent wage)

― Karl Malone, Thursday, 19 January 2023 bookmarkflaglink

Or it's everyone being "paired" with an AI. So you have a editor paired with an AI, or a paired programmer. The point is to downgrade the value of what people can do, and therefore downgrade their wages.

As bad as a lot of AI art is, CrAIyon is amongst the worst at making images.

xposts - I agree, the USP of something like Craiyon is the inspirational randomness, the uncanny parts, not the final product

I think ChatGPT (or a version of it) has serious worth for students who struggle with essay-writing, helping them craft arguments and structure paragraphs of writing; it's a good writing tutor, if you can resist the temptation to copy-and-paste

sorry, "paragraphs of writing" is a horrible phrase, should've used a chatbot

those horizontal lines on the left and his rectangle arm.

that's just a brick wall, no? prompted by "pink floyd" I assume

current GPT is way too error-prone to let students use it unsupervised imo

"Joan, Maximilien Robespierre was not the original drummer for the Beatles. F-"

https://edition.cnn.com/2023/01/16/tech/facial-recognition-fashion/index.html

"Didero, 29, who’s studying for a PhD ... says the idea for Cap_able came to her when she was on a Masters exchange at the Fashion Institute of Technology in New York."

alternatively she could've read Zero History by william gibson from 2010

I was arguing with chat gpt about what was Ed Wood's final film - I think it won

Source: https://t.co/HpAO2lQ6zm

— Genevieve Roch-Decter, CFA (@GRDecter) January 26, 2023

for now, it's just for "making more comprehensive quizzes, interactive content"

the market seems pleased by this efficiency

i'll show myself out
https://drayk.it/song/ac9fb806ce86493eb3bdf35224100115

https://google-research.github.io/seanet/musiclm/examples/

interesting

lol forks

https://drayk.it/song/7511f00ee39e40c78919f4e84d3fcc8c

the ai songs latham posted are not good per se, but they're eerily good for some ai generated bullshit. the tempo seems really off for a lot of them though.

https://drayk.it/song/690478d82f424198a22e5ab54724d5ae

ChatGPT's Ian Curtis lyrics were poor - but its Bernard Sumner lyrics were very plausible indeed

the MusicLM stuff is interesting but you can definitely tell it's AI, it's full of audio sludge and it sometimes hits the space that's like between two notes. that said some of this sounds pretty modern to the point where I kinda wonder if this technology has actually been around for a while. like it would not surprise me to find out all those Swedish songwriting teams have been using something like this for years and just curating the best parts

The hybrid or “centaur “ approach to ai as a tool rather than human replacement

ChatGPT's Ian Curtis lyrics were poor - but its Bernard Sumner lyrics were very plausible indeed

― Zelda Zonk, Monday, January 30, 2023 10:58 PM (yesterday) bookmarkflaglink

Haha well I mean....

actually really good

https://www.twitch.tv/watchmeforever

you had to be there but this was a moment

https://i.imgur.com/f4okuVh.png

Every Sumner lyric involves some awkward bit of inadvertent crashing bathos, this is probably ideal for an AI.

making characters look directly at camera pic.twitter.com/Yl9hG31Ae9

— ActionMovieDad (@ActionMovieKid) January 24, 2023

alright this is kinda freaky (also very funny)

actually really good

https://www.twitch.tv/watchmeforever🕸