Dall-e is a statistical model for visual data. It makes visual data that looks like stuff

Chatgpt is a statistical model for text data. It makes text that looks like stuff. That's all. You can't rely on it to produce text that's factually accurate, or code that works, because that's not what it does

That doesn't mean it's useless. When you have to come up with words for something that "look right", like when you're writing an email or ad copy or a self description blurb for a resume, it can generate something for you that looks like the words you need. And for that it works great

@eniko I am so fed up with seeing “look, chatgpt can do X!” the latest X I saw being “generate crochet patterns”

No, it just slurped up (stole) a bunch of random crochet patterns and now it can just throw bits of them together at random, or if it’s really lucky you’ll ask for a pattern it already has and can just throw back at you

I am begging people to learn what this thing actually is before getting so enthusiastic

@eniko , The only AI that I consider useful and the ones that I use are the search engines and not precisely those of Google or Bing. In my opinion, especially recommendable 2, Andisearch, was the first search engine that included AI, before the two big ones, and recently also Perplexity, based on OpenAI. Both for anonymous use without registration, without ads, trackers, logs or other garbage.

https://andisearch.com

https://www.perplexity.ai

See also

https://alternativeto.net/news/2023/2/chatgpt-and-privacy-why-you-shouldn-t-trust-ai-with-your-secrets/

@eniko the level of fear is ridiculous. Its kinda sad too. Theres so much cool stuff happening with ai tech but people are all too busy panicking over all these scenarios they’ve made up in their heads.

I wished the models were better though. The only language model that seems capable of generating somewhat usable code is chatgpt but im not as excited about it because its basically going to be a proprietary service.

@eniko @browren

1. You say ChatGPT is not AI. OpenAI, the makers of ChatGPT, say it is.

2. The essay in question deals with non-maleficence, which is an attribute that does not require agency, let alone actual intelligence.

@eniko

This is a good point. The public story around ChatGPT seems to treat it as an alive entity when in reality it is just a model.

It's just inputs and outputs. It's going to give you back what it's programmed to. There is no originality or thought here, it's emergent behavior based on all of its inputs and whatever math is happening under the hood.

@eniko
Thisy this this this.

I've been having an argument in another thread, but can't seem to get it through people's heads: calling it "AI" is not just an issue of semantics, it's got deep ramifications for how we treat it, especially those that don't understand it.

(I am thinking of calling it 'metaverse+' just to try to kill it)

@eniko This entire affair is a perfect example of why "a lie is halfway round the world before the truth has got its boots on"

A lie explains, justifies and propagates itself. The truth has to actually educate people. It's going to be a real struggle getting a critical mass of people to understand how bad ChatGPT is.

@eniko

Fake Intelligence.

Something that is intended to convince people there is some form of intelligence but there isn't.

@eniko

You are technically, correct, which is the worst kind of correct. You aren't going to be able to educate the general public, even moderately intelligent individuals. The perception is that it IS an AI. And people will base decisions off it it.

It doesn't matter what definitions it does or does not fit, because letting this horribly broken thing into the wild will lead to increase in harm (to both individuals and society).

@eniko Gods this though. Like, there's a certain amount of anthropomorphization going on, and among other things, it's allowing people to forget what they're looking at, what to expect, how much human labor goes on behind the scenes, and who is "responsible," so to speak, for its behavior.

I've had some frustrating conversations, if that wasn't readily apparent by the above half-rant.

@eniko I keep saying that it's amazing how quickly the world went from "hey, ChatGPT can tell a decently funny story, even if it doesn't always stick the landing" to "OMG ChatGPT GAVE ME BAD ADVICE ABOUT MY MEDICAL PROBLEMS!"

And now I wonder how much attention I could get by complaining that Dall-E provides worthless images when asked for security footage of my front door...