Artificial intelligence still has some way to go

v behind on this thread, but anyone know any good writing on models for describing GenAI in terms of information theory/cybernetics? i’m reading ruyer on cybernetics, mainly looking at the problem of the origin of information in cybernetic theory, and while my instinct is that genAI *can* be captured/described by cybernetic theory, i feel GenAI does require such a description or some additional work at the margin (ie it can’t be captured entirely by pre-existing thinking).

king ludd (has surely come)

https://www.theautopian.com/a-mob-just-vandalized-and-set-a-waymo-self-driving-car-on-fire-and-the-videos-are-nuts/

Hell yeah

It’s not clear who exactly attacked this Waymo in Chinatown on the Lunar New Year, or what their motivation was...

Just spit-balling here, but could it be that self-driving cars are a highly visible symbol of Silicon Valley's wealth and political ascendancy over the Bay area in general and SF in particular and their presence on the streets isn't just a constant reminder of this, but is perceived by people on the street as a danger and an insult imposed on them by those who possess that wealth and power? Or maybe it was just a mysterious, aberrant and irrational act and we'll never know what possessed that mob to do that.

the guy who wrote that piece/runs that blog has a tongue coated in boots

i enjoyed his evident unease

my favorite part is where the commenters liken the people desecrating the vehicle to “rapists” and “torturers” lmfao

Lol did someone fuck the car

I remember reading an Ebert review of Crash (the Ballard one) which defended its non-porn status on the basis that no one would actually want to fuck a car but uhm...

“hey elon, the peasants are revolting!”

in the age of the internet it is no longer possible to believe that there aren't people who want to fuck cars

in the age of ai and crispr or whatever i have no doubt someone to actually go it and produce horrifying offspring or allegations. virtually of course.

“go and do it” christ i cannot even type unfunny shit right. but this was serious.

I’ve seen Crash and Titane and gone to a car show or too. “Muscle” cars hmm yeah I see you there, buddy

oh wait, forgot The Counselor (director’s cut of course)

https://us.rule34.xxx//images/655/6528d0aa955fda17fe8773954c3f7a2bef943eb8.png

hey now I’m not a pervert

This is beautiful, to me.

https://www.wired.com/story/confessions-of-an-ai-clickbait-kingpin/

lmao this is great

https://www.columbiamissourian.com/news/state_news/missouri-appeals-court-fines-litigant-after-finding-fake-ai-generated-cases-cited-in-filings/article_e4cc108a-cac1-11ee-b9ca-2fa1b1765ce7.html

Awesome, much respect

Well you've got to give him all of the respect due to any entrepreneur who's found a niche market and exploted it (that is absolutely zero respect)

imo the core quote of that Wired story xyzzzz linked:

A plum domain’s initial benefit—a strong reputation with Google and a built-in audience—dwindles quickly as Vujo populates it with content primarily designed to snare search engines rather than interest people. AI content is successful not because it is replacing the work of human writers but because it coasts on the value created by their past labor.

Private equity firms like Bain Capital have long been using the same basic strategy as that Serbian clickbait guy. They buy companies with classic established brand names associated with reliable quality, degrade the products heavily, cut prices moderately, revive profits based on gulling customers into thinking they are getting the same quality the brand always delivered in the past, but cheaper, then dump the company before the public gets wise to the changes. Main difference is that the web sites he buys are small potatoes. They don't even have pension funds to raid.

I saw a job listing at LinkedIn for people to take photos per their assignment, with each one that is accepted earning them $1. The company is apparently building a database of images to use for AI, with no copyright issues.

I'm not saying Generative AI will never pass for real but the fact some people boosting it now think this comedic shit makes him "feel like he's in the room and can smell the candles" and "reminds him of his own grandmother" makes me wonder

Video #3

Prompt: A grandmother with neatly combed grey hair stands behind a colorful birthday cake with numerous candles at a wood dining room table, expression is one of pure joy and happiness, with a happy glow in her eye.

She leans forward and blows out the candles with a… pic.twitter.com/3M4Ey3503t

— GREG ISENBERG (@gregisenberg) February 15, 2024

i think MR James would love this

When designer aims for 'fluid UI' but takes it too literally.#sora pic.twitter.com/cUPt9vARjk

— Yasir (●ᴗ●) (@yasirbugra) February 15, 2024

Oh my god, the cake video is terrifying

As someone says in the comments, it's very old-couple-laughing-maniacally-in-Mulholland-Drive. And the hands of the woman next to her!!

The table bisecting the background dude on the right is also disconcerting

Far left lady's clapping is ... wrong

After months of resisting, Air Canada was forced to give a partial refund to a grieving passenger who was misled by an airline chatbot inaccurately explaining the airline's bereavement travel policy.

On the day Jake Moffatt's grandmother died, Moffat immediately visited Air Canada's website to book a flight from Vancouver to Toronto. Unsure of how Air Canada's bereavement rates worked, Moffatt asked Air Canada's chatbot to explain.

The chatbot provided inaccurate information, encouraging Moffatt to book a flight immediately and then request a refund within 90 days. In reality, Air Canada's policy explicitly stated that the airline will not provide refunds for bereavement travel after the flight is booked. Moffatt dutifully attempted to follow the chatbot's advice and request a refund but was shocked that the request was rejected.

Moffatt tried for months to convince Air Canada that a refund was owed, sharing a screenshot from the chatbot that clearly claimed:

If you need to travel immediately or have already travelled and would like to submit your ticket for a reduced bereavement rate, kindly do so within 90 days of the date your ticket was issued by completing our Ticket Refund Application form.

Unhappy with this resolution, Moffatt refused the coupon and filed a small claims complaint in Canada's Civil Resolution Tribunal.

https://arstechnica.com/tech-policy/2024/02/air-canada-must-honor-refund-policy-invented-by-airlines-chatbot/

That chair video has Rubber vibes. Chair (The Chair That Kills People) coming soon from A24

there were a spate of Twitter posts, maybe mentioned upthread, where people had got chat bots to sell them cars for $1. i wonder if those will also hold up in court.

I just bought a 2024 Chevy Tahoe for $1. pic.twitter.com/aq4wDitvQW

— Chris Bakke (@ChrisJBakke) December 17, 2023

That grandma clip looks like Chris Cunningham made it for an Aphex Twin video

Jack the Ripper’s face finally revealed — here’s what AI thinks the notorious killer looked like https://t.co/IIpl4pewOW pic.twitter.com/zpDx8UcdTc

— New York Post (@nypost) February 17, 2024

Prepared to grudgingly accept AI if it means we lock away Johnny Depp

harnessing the power of machine learning to reveal saucy jack was hot stalin all along

https://i.stack.imgur.com/qDbU1.jpg

Science fiction writers: The legal case for robot personhood will be made when a robot goes on trial for murder.
Reality: The legal case for robot personhood will be made when an airline wants to get out of paying a refund. https://t.co/aTGdErEr9g pic.twitter.com/4JiVLZRhq2

— Chris Farnell (@thebrainofchris) February 17, 2024

There is precedent!

https://en.wikipedia.org/wiki/Deodand

mission: make robots actually fear own death

side effect: creation of procreationql imperative and impulse.

solution: imprisonment and capital punishment.

hey i’m only human.

born to— nah

they should’ve just said that the chatbot that provided the wrong advice has been.. terminated

i'm deodanding better than I ever did

https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/af79fdb5-5696-4095-bb60-682d45bcd45c/Screenshot_2024-02-19_at_2.52.45_PM.png?t=1708365198

an agile Indian police officer with really weird hands...

that's... a real movie

https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai/

I'm interested in this "data poisoning tool" - could it be easily overridden?

I can't imagine the AI people won't find a way around data poisoning as it currently is, but if the data poisoning people keep at it and keep finding new ways to poison the data, it could become an interesting arms race.

If I understand their aim correctly, it’s obfuscation. I assume it’ll work in the sense that nobody will bother to mitigate the obfuscation. Nevertheless, I don’t think it matters from a “foundational model” perspective, unless model architectures dramatically change, as performance has been seeing diminishing returns from increasingly large dataset sizes.

there's a twitter link uothread 3 weeks ago saying nightshade is easily defeated.