Artificial intelligence still has some way to go

Message Bookmarked
Bookmark Removed
Not all messages are displayed: show all messages (8884 of them)

THE CEO OF MAZDA JUST SAID THE 2025 CX-5 IS GONNA BE THE BEST ALL AROUND VEHICLE ON THE MARKET AND EVERYONE IS ACTING LIKE THIS IS NORMAL https://t.co/hhXrSd7QZ9

— Jesse Brenneman (@Jesse_Brenneman) September 24, 2024

𝔠𝔞𝔱𝔹 (caek), Wednesday, 25 September 2024 16:40 (one year ago)

lol

Ʉɯ ïž” (°□°) (mh), Wednesday, 25 September 2024 16:45 (one year ago)

Fwiw if yall are in the market for some cool shirts I discovered this brand while I was in Malaysia and they make quality stuff, they ship worldwide via DHL and the stuff shows up so much faster than you’d think

https://www.kaptenbatik.com.my

trm (tombotomod), Wednesday, 25 September 2024 21:08 (one year ago)

openai decided its not a non profit anymore and gave sam altman $150b in stock then a bunch of their top people resigned

lag∞n, Thursday, 26 September 2024 00:33 (one year ago)

really they gave him $7b but you get the idea

lag∞n, Thursday, 26 September 2024 01:02 (one year ago)

https://pbs.twimg.com/media/GYQWRhrWgAIL2C3?format=jpg&name=medium

lag∞n, Thursday, 26 September 2024 01:29 (one year ago)

cloud-based mouthfeel

mookieproof, Thursday, 26 September 2024 01:30 (one year ago)

Senate Foreign Relations Committee Chair Ben Cardin (D-Md.) was recently the target of a sophisticated deepfake operation impersonating a top Ukrainian official, according to three people briefed on the matter and communications reviewed by Punchbowl News.

...According to the notice, the individual was impersonating the recently-former Ukrainian Foreign Minister Dmytro Kuleba. “Kuleba” was asking questions during the call that the participants thought were bizarre, including: “Do you support long-range missiles into Russian territory? I need to know your answer,” according to the notice sent to senior Senate aides.

The impersonator was speaking in the voice of Kuleba, which sources said they believed was likely recreated using artificial intelligence. The notice stated that the impersonation effort had “technical sophistication and believability.”

The notice sent to top Senate officials added that the individual was asking “politically charged questions in relation to the upcoming election
 likely trying to bait the senator into commenting on a political candidate.”

https://punchbowl.news/archive/cardin_ukraine_deepfake/

z_tbd, Thursday, 26 September 2024 16:01 (one year ago)

suggestion for deepfake tinkering: add "i need to know your answer" to the "do not say" list

z_tbd, Thursday, 26 September 2024 16:02 (one year ago)

When Mr. Altman visited TSMC’s headquarters in Taiwan shortly after he started his fund-raising effort, he told its executives that it would take $7 trillion and many years to build 36 semiconductor plants and additional data centers to fulfill his vision, two people briefed on the conversation said. It was his first visit to one of the multibillion-dollar plants.

TSMC’s executives found the idea so absurd that they took to calling Mr. Altman a “podcasting bro,” one of these people said. Adding just a few more chip-making plants, much less 36, was incredibly risky because of the money involved.

https://archive.ph/dglmB#selection-1149.0-1153.234

lag∞n, Thursday, 26 September 2024 20:33 (one year ago)

A VC firm I had a Zoom meeting with used Otter AI to record the call, and after the meeting, it automatically emailed me the transcript, including hours of their private conversations afterward, where they discussed intimate, confidential details about their business.

— Alex Bilzerian (@alexbilz) September 26, 2024

lag∞n, Friday, 27 September 2024 02:12 (one year ago)

Oopsie

Muad'Doob (Moodles), Friday, 27 September 2024 02:32 (one year ago)

whom amongst has not forgotten to end a call, let alone software recording that software?

(really though, Teams is really bad at this because if the recording party forgets to close the call it has a single-participant nothing-happening recording hours past the end)

Ʉɯ ïž” (°□°) (mh), Friday, 27 September 2024 03:34 (one year ago)

nightmare material

Humanitarian Pause (Tracer Hand), Friday, 27 September 2024 08:53 (one year ago)

good Max article on the preponderance of AI slop: https://nymag.com/intelligencer/article/ai-generated-content-internet-online-slop-spam.html

jaymc, Friday, 27 September 2024 13:06 (one year ago)

Just reading that.

"Fifteen years ago, Wired magazine heralded the “good-enough revolution” in low-cost technology: “Cheap, fast, simple tools are suddenly everywhere 
 We now favor flexibility over high fidelity, convenience over features, quick and dirty over slow and polished.” Generative AI as a technology exists in this lineage. That it can create adequate texts and images is an astonishing leap forward in machine learning, but the texts and images are still only adequate, “good enough” and cheap enough for people to thumb past on their phones. Slop is the most appropriate word for what it produces because, as disgusting and unappetizing as it may seem, we still eat it. It’s what’s right there in the trough."

It was always the way.

xyzzzz__, Friday, 27 September 2024 16:56 (one year ago)

It's just applying that old rule of thumb that if keeping your standards up costs money or effort, just your standards.

more difficult than I look (Aimless), Friday, 27 September 2024 17:02 (one year ago)

dammit! just lower you standards

more difficult than I look (Aimless), Friday, 27 September 2024 17:03 (one year ago)

max is currently sifting through a pile of emails that belittle him for thinking that anyone would ever be fooled by ai

z_tbd, Friday, 27 September 2024 17:09 (one year ago)

worse but cheaper is the original idea of business disruption, consept being that the upstarts capture the low end of the market first and the big players dont really care cause its not that profitable but then the upstarts use that as a beach head to take the rest of the market, not sure if that totally applies to AI but it does resemble it more than a lot of tech stuff that talks about disruption in a more generic way

lag∞n, Friday, 27 September 2024 17:20 (one year ago)

Offshore Labour, marketing scams, plagiarism, overproduction of content -- whether its posts, genre fiction, academic papers -- and all of this so we can get the pennies we need to make rent next month. This so is in step with the world we live in, slop is fully attuned to our reality xp

xyzzzz__, Friday, 27 September 2024 17:21 (one year ago)

were def living in slop times

lag∞n, Friday, 27 September 2024 17:22 (one year ago)

"AI-generated papers, Cabanac argued, are often used to pad an academic’s rĂ©sumĂ© with more publications and citations: “You buy a paper on a topic of your choice, and you buy a set of 500 citations. Then you go to your faculty and you say, ‘Look, I’m a genius, and I deserve this position as a full professor.’” In other words, like Facebook slop, the content of the content isn’t really as important as its presence — or, more accurately, its measurability."

^ If you look at this, slop and AI are the least of our problems.

xyzzzz__, Friday, 27 September 2024 17:26 (one year ago)

So go to https://notebooklm.google.com/ and start feeding it YouTube videos, audio files, documents, PDFs, anything... and it will produce summaries, table of contents, and study guides for the whole aggregation - then choose Audio Overview and it will generate a 6-10 minute conversation between two unnamed hosts discussing it as if it were NPR taking a deep dive.

So I fed it the entire July 2024 US Politics thread and this was the result. A couple minutes in you can tell it's AI generated slop but when they start mentioning ILXors by name it's startling.

Elvis Telecom, Saturday, 28 September 2024 03:36 (one year ago)

you can tell it's AI generated slop

all those incisive generalities and cliche-ridden observations! almost like the real thing, but much worse.

more difficult than I look (Aimless), Saturday, 28 September 2024 04:01 (one year ago)

Slop is often fine in its way - feeding a prompt into Midjourney and seeing what comes out is fun, if it's something reasonably interesting you can share it with other people who are also amused by AI image generation shitposting. There is theoretically an outlier where someone could generate meaningful art with it though I have not seen it (I was doing photography and constructivism in college, I'm not allowed to throw stones at art forms based on ease of creation).

Too bad about capitalism and the drive to destroy the climate ruining all of it.

papal hotwife (milo z), Saturday, 28 September 2024 08:31 (one year ago)

So I fed it the entire July 2024 US Politics thread and this was the result. A couple minutes in you can tell it's AI generated slop but when they start mentioning ILXors by name it's startling.

lol the hosts are major frogbs stans

Daniel_Rf, Saturday, 28 September 2024 08:34 (one year ago)

I think when Trump wins that might be better than the US pol thread tbh.

xyzzzz__, Saturday, 28 September 2024 10:59 (one year ago)

oh no paraphrasing unperson as saying "wake up people" is too mean

rob, Saturday, 28 September 2024 14:20 (one year ago)

A few hallucinations:

1. "President Keyes said, 'Diet Mountain Dew is racist? What is even happening anymore?'

This is close to an actual post, but it's from eephus, not President Keyes ("Now I'm like -- Diet Mountain Dew is racist? Harris is bad because she laughs? This makes sense to someone -- but who???")

2. "collardio gelatinous even compared it to that whole santorum thing."

It was Ye Mad Puffin who brought up "santorum."

3. "unperson said 'Focusing on the ick factor, for lack of a better word, could be really effective, especially for voters who are already turned off by the Republican Party."

unperson does have posts with this general sentiment, but not in these exact words. The term "ick factor" does not appear in the thread at all.

jaymc, Saturday, 28 September 2024 15:11 (one year ago)

(I'm sure there are more, those are just the ones I checked.)

jaymc, Saturday, 28 September 2024 15:12 (one year ago)

lmao it made a fn podcast this is some twisted shit

lag∞n, Saturday, 28 September 2024 15:15 (one year ago)

unsurprisingly my posts were too radical for AI

rob, Saturday, 28 September 2024 15:17 (one year ago)

theyre going off about how powerful an image of trump fist bumping a supporter is

lag∞n, Saturday, 28 September 2024 15:19 (one year ago)

wait til you get to the rizzler (?) part

rob, Saturday, 28 September 2024 15:20 (one year ago)

da rizzler baby

lag∞n, Saturday, 28 September 2024 15:21 (one year ago)

goldfish podcasts hosts moving from topic to topic all omg so true, a summary presented in the style of analysis

lag∞n, Saturday, 28 September 2024 15:24 (one year ago)

Is there any practical application for this LLM slop other than corporate pricks spewing a tide of word-salad garbage into the world so they can sack all their employees? Or various other pointless uses. I'm sure this has been discussed in the thread but time is money, right?

RIO Speedwagon (Matt #2), Saturday, 28 September 2024 15:25 (one year ago)

its interesting that it knows things about the topic it presumably has been fed a bunch of election news, wonder how it would do with more obscure or abstract topics

lag∞n, Saturday, 28 September 2024 15:26 (one year ago)

if tech like this could say summarize a meeting and actually pull out the important parts so one could listen to a shorter version then it would be useful but it doesnt know anything about importance

lag∞n, Saturday, 28 September 2024 15:43 (one year ago)

ai summarized a 30 minute doctor visit i had a couple months ago. the doctor showed me the results and let me read through it all. it was actually crazy accurate, and considering that in the last 10 years i've never seen my primary care physician more than once (every single time, they changed hospitals/insurance plans before i could see them a second time), i thought it was an improvement

z_tbd, Saturday, 28 September 2024 15:44 (one year ago)

something like that where its completed technical info makes sense, theres no need to understand anything, similar to how its useful for c0ding

lag∞n, Saturday, 28 September 2024 15:47 (one year ago)

tho its certainly not trustworthy

lag∞n, Saturday, 28 September 2024 15:49 (one year ago)

so like maybe useful for a presentation tho still requiring verification, not really useful for a discussion

lag∞n, Saturday, 28 September 2024 15:51 (one year ago)

verification step is obvs going to get skipped or done halfheartedly a lot leading to undesirable outcomes

lag∞n, Saturday, 28 September 2024 15:58 (one year ago)

https://www.tiktok.com/@onytitned/video/7416057391722089761

韜, Saturday, 28 September 2024 16:50 (one year ago)

but also it sucks when stuff like this uses AI

https://www.amny.com/wp-content/uploads/2024/09/unnamed.png

i think the big bagel makes enough money from bagels to throw a couple hundred towards a designer for a poster design that doesn't suck

韜, Saturday, 28 September 2024 17:00 (one year ago)

Burger King logo gonna sue.

pplains, Saturday, 28 September 2024 18:03 (one year ago)

lol thats so bad

lag∞n, Saturday, 28 September 2024 18:15 (one year ago)

the American flags are killing me

jaymc, Saturday, 28 September 2024 19:01 (one year ago)


You must be logged in to post. Please either login here, or if you are not registered, you may register here.