Artificial intelligence still has some way to go

they still havent said wtf he did they accused him of lying what was it tell us

i thought it was basically that he said he supported guardrails and safer advancement of the research but then turned around and did pushed for faster development

"did pushed" sorry this post was written by chatGPT. "did the exact opposite"

i mean prob basically but they came out and publicly accused the guy so they owe it to us the rubberneckers to formally break it down in detail

They'll just place an classified ad in the Personal section of the SF Chronicle saying: "Sam, come home. All is forgiven. We kept your room exactly as you left it. We miss you."

i think it had something to do with ai safety. sam altman was bullish, the board was more reserved.

yup my best rubbernecker guess is that something about Altman's commercial big plans (maybe especially 'starting a chip fab plant specifically for AI chips (using Saudi money)'?) tipped them over into 'no no no this is not nice slow safe progress'.

There's something about chatgpt's remorseless cheerful will to mediocrity that I cannot look away from.

User
Hi Chat GPT. I've started a band and I want it to be a mix between The Stone Roses and Primal Scream with the swagger of Oasis. Can you suggest some names for my band?
ChatGPT
Starting a band with a blend of The Stone Roses, Primal Scream, and the swagger of Oasis sounds exciting! Here are some band name suggestions that might capture the essence of these influences:

Rose Haze Revival
Electric Scream Parade
Swaggering Roses
Primal Oasis Echo
Stone Vibe Syndicate
Echoes of Madchester
Britpop Rebels
The Neon Stone Echo
Scream Oasis Fusion
Retro Sonic Bloom
Each of these names tries to encapsulate elements from the distinct sounds and attitudes of the bands you're inspired by. Feel free to mix and match words or elements from these suggestions to create something that feels right for your band!

User
We're not a tribute act
ChatGPT
Understood, you're looking for a unique identity while drawing inspiration from those bands. Here are some original band name ideas that capture the essence of your influences without directly referencing them:

Velvet Echoes
Cityscape Serenade
Midnight Brit Groove
Indie Vortex
Rising Tempests
Urban Mythos
Rebel Hymn Harmony
Skyline Drifters
Neon Anthems
Ethereal Brigade
These names aim to evoke the spirit of the Britpop and indie rock scene while maintaining a distinct identity for your band. They combine elements of urban, rebellious, and musical themes to reflect a modern, energetic sound.

LLMs cannot be otherwise. They “represent the norm by signaling the mean” to quote hito steyrl. You’re just seeing a model of averages in the data.

Even when you get “weird” results it is due to the LLM misunderstanding the question. They cannot produce a truly unique response to a prompt, one based on insight, because they do not understand it.

A lot of speech that goes on in media and business is mimetic for sure — people reacting in more or less predictable ways to situations. But there is always the chance that a person might develop a new understanding of something, or take a different approach. The LLM cannot do this so it is not a replacement for a human being imo. I don’t think they’ll even ever really do customer service, because that requires responding to novel situations every once in a while.

Treeship that explanation makes a lot of sense to a philistine like me. Ty

I am also very disappointed in those band name suggestions. Come on, man.

I don't disagree in principle - certainly that human intervention is currently needed, a lot - but I'm a bit more pessimistic about how much of the world of work straight statistical mimesis can eat, especially as gen ai improves.

they still havent said wtf he did they accused him of lying what was it tell us

― lag∞n, Monday, November 20, 2023 6:42 PM (yesterday) bookmarkflaglink

Sutskever is said to have offered two explanations he purportedly received from the board, according to one of the people familiar. One explanation was that Altman was said to have given two people at OpenAI the same project.

The other was that Altman allegedly gave two board members different opinions about a member of personnel. An OpenAI spokesperson did not respond to requests for comment.

https://www.businessinsider.com/openais-employees-given-explanations-why-sam-altman-out-2023-11

We've seen decades of offshoring and automation and yet there are lots of people needed to do work. It's just whether there is the will to invest in the infrastructure around it to support it.

In the area I know a bit about a machine doing some of the heavy lifting around coding isn't the issue. I mean, offshoring has already taken out a lot of the cost anyway.

But can AI help define and solve the business problems?

an excerpt here about how altman previously being a fuckup https://www.newcomer.co/p/sam-altman-forced-out-of-openai-by

people will still lose jobs. businesses cut costs in many ways that hurt their product. my point is that AI is overhyped. the labor power of human workers is still the engine of capitalism, and people who say otherwise are lying.

i just do not believe this technology is that revolutionary. there have been chatbots for decades. these are more uncanny. but the way they are generating their outputs does not resemble the way human intelligence works. it's sophisticated data modeling.

as a bandname ETHEREAL BRIGADE is in fact good not bad

all the others suck tho

lol PRIMAL OASIS ECHO

Treeship speaks wisdom

people will still lose jobs. businesses cut costs in many ways that hurt their product. my point is that AI is overhyped. the labor power of human workers is still the engine of capitalism, and people who say otherwise are lying.

Here's the thing tho: "generative" (eyeroll) AI is only ever recycling amd recombining existing content. It has never "generated" anything or come up with an original thought. Its only fuel is what people (people like us) have already given it.

I have spent, um, something like 38 years feeding it. Every word that I have published, every illustration I've drawn, every song I have written and recorded.

The robots have not thanked me, as far as I know. And when I am unemployed, probably next month, they will not pay my mortgage. So forgive me if I am little bit bitter about that.

xps

I still lean to it being revolutionary, at least for laptop jobs - it's come along quickly, it's still developing fast, even in current state its possibilities when trained on specialist datasets are just getting started. The 'not like human intelligence' feels a bit marshlight to me - it doesn't really need to be to cause a lot of disruption in the near future, and if it develops rapidly I'm not sure it matters whether its mode of 'intelligence' is recognisably human or instead non-human - we'll be somewhere messy either way.

I can definitely see a path with medium-term stability where it's just another tool, as it is now, but I'd feel naive banking on that (because everything always gets worse).

YMP you're welcome to join my new band, Midnight Brit Groove

I mean maybe as someone who's made a living generating sentences - plenty of them pretty rote - I am just falling for the uncanniness. I know how dumb it is! eg how bad at mimicking styles where it has a narrower dataset.

One of my bosses has started to develop the annoying trait of searching for information of OUR OWN PROCESSES in chatgpt, a clear indication of what we were all suspecting in that he knows sod all about the job.

an excerpt here about how altman previously being a fuckup https://www.newcomer.co/p/sam-altman-forced-out-of-openai-by

― 龜, Tuesday, November 21, 2023 8:27 AM (one hour ago) bookmarkflaglink

thats the one i was talking about upthread

how many swords, Dall-E, how many swords?
https://imgur.com/a/e5VgCCm

chatgpt et al absolutely need to have a huge banner that shows up every session saying NOT A SEARCH ENGINE! I MAKE SHIT UP!

We're def going to see a lot more from the entertaining genre of 'people fucking up really badly because they trust ChatGPT as some kind of god-mind'

chatgpt et al absolutely need to have a huge banner that shows up every session saying NOT A SEARCH ENGINE! I MAKE SHIT UP!

― organ doner (ledge), Tuesday, November 21, 2023 9:58 AM (ten minutes ago) bookmarkflaglink

now why would they do that lol

sorry, need to dial down my faith in humanity settings

just went to check out bings situation and its interface features the ai bot interface heavily tho if you type into the box itll take you to normal search results with "chat" being a tab, btw lol at the prompt

https://i.imgur.com/t0nvuXA.png

chatgpt isnt a search engine doesnt have access to new info but microsofts implementation of it is?

https://i.imgur.com/0dLTvM3.png

no, it's got integrated bing search now

training cut off is April 2023 tho

so I think if you say 'tell me about Bjork' you get a summary of her life and works to 4/2023

But

User
What's Bjork's latest release?

ChatGPT
Björk's latest release is a new dancehall-inspired song titled "oral," a collaboration with Rosalía. This track was officially released after being teased earlier, and all profits from it will go towards fighting fish farming in Iceland

And it says 'researching with bing' while it's generating.

so it may or may not be a search engine depending on what you ask. Can't see any trouble there

no, it's got integrated bing search now

― woof, Tuesday, November 21, 2023 11:49 AM (seven minutes ago) bookmarkflaglink

but what does this mean

I want to live in this world

(What happened to post notifications? That was supposed to be in response to this frankly amazing-sounding Björk song)

microsoft’s a licensee and their bing product using chatgpt is what I mentioned earlier: a derivative that was fed bing’s search data and was instructed to return verbatim data and not fully generative phrases

I know a few people working with the Microsoft-hosted white label ChatGPT to get it to serve up factual data and it’s tricky. It’s very particular in how it reads structured data and one of the possibilities would be not to point it at internal wiki-style articles that are vary wildly in structure and format, but to write documentation in a very structured format, and then have the wiki-style stuff be downstream from the LLM. So you’d write very concise statement/description paragraphs, feed those to the chatbot for q&a purposes, and also have chatgpt generate the pages.

Obviously you’d need proofreading and this is all very beta, but I think we’re at yet another iteration of restructuring how we write in order for computers to process that information and then deliver it to humans.
This has, of course, gone swimmingly with SEO and clickbait in the past

if you’re wondering what the product looks like: https://learn.microsoft.com/en-us/azure/ai-services/openai/use-your-data-quickstart

My professional world is largely technical writing, in a competitive context.

An AI that goes out to the internet and does writing-by-google is worse than useless, because every competing company can submit exactly the same content.

The tools that I think have more potential are those that do NOT go outside, and just parse my existing library. But... I could just do that myself.

I think the risk isn’t that human writers go away, it’s that all writing becomes this consumed-by-ai technical writing specifically structured in ways that make sure the LLM doesn’t fuck it up when it regurgitates it

chatgpt gets extra functionality quite often - now it does pictures, now it reads PDFs, now a bit of data viz, now it searches the web. And sometimes a bit of it goes, and then it comes back… it's a confusing platform tbh

The MS thing sounds like it overlaps with GPTs?

yes. they’re a close licensee of chatgpt, kind of a second-party provider, and possibly on the way to being a licensee that actually does most of the development as part of a consortium

I think the cynical take I've landed on is that the product of AI will kinda remain half-assed garbage but that people will slowly adjust to the garbage until it becomes an acceptable alternative to human-generated product in many contexts. Like I can see the writing staff of every show on CBS being replaced by AI without anyone really noticing or caring.

fwiw human language is only one of the things LLMs are decent at. they’re basically pattern recognition and inference tools

https://arstechnica.com/science/2023/03/large-language-models-also-work-for-protein-structures/

afaict one of the teams I support is using a sieve-like approach to run multiple LLMs and other ML models to infer protein structures in the way mentioned in this article

xp maybe! I think the crass commercialization aspect is that we’re going to get more sophisticated versions of chum. those nonsensical chumbox articles at the bottom of pages that Clickhole satirizes will start looking different. episodic tv of the NCIS or w/e variety is basically background noise chum already and it’s easy to imagine that being generated

the issue is that there are no new ideas and you still have to have something to pull from, whether it’s a decade of scripts or just throwing all of the past tv shows in a blender plus something new. to an extent all human endeavors are a mishmash of existing material but it’s disarming to think that the AI could eventually build new stories off of the AI’s previous output

i dont think this thing will be able to write a tv show but maybe it will be able to write some of the stuff no one cares about

https://i.imgur.com/LTZ9dGM.png