Good Morning Post

[Cranesbill]

everything has uncertainty

it doesn't mean we should stifle innovation

I bet you can't even tell I made this with AI

View attachment 570437

Not all uncertainty is equal

There’s a difference between tolerable risk (like putting on new clothes) and high-impact uncertainty (Like systems that shape education, healthcare, or public information)

And AI can't generate copyright characters or get lighting right. You drew this

 

[AncientBoi]

uhh, what place?

 

[Cranesbill]

 

[x65943]

Not all uncertainty is equal

There’s a difference between tolerable risk (like putting on new clothes) and high-impact uncertainty (Like systems that shape education, healthcare, or public information)

And AI can't generate copyright characters or get lighting right. You drew this

AI 100% can generate copyrighted characters wut

 

[Cranesbill]

AI 100% can generate copyrighted characters wut

But it's never 100% accurate

 

[x65943]

But it's never 100% accurate

just admit you are jealous of my AI skills

 

[Cranesbill]

Ew

@hippy dave can you help me out here? This is like the one thing we have in common

 

[x65943]

AI is getting better every day

last time I asked it to make that on the right, it said it was too sexy for its servers to comprehend

 

[x65943]

 

[AncientBoi]

Alas, poor coffee shop

 

[kijetesantakalu042]

For once I agree with Crane here.

Generative AI (excluding SynthV/Vocaloid 6) is useless, and the awful data centers are making everything worse. I'd hate to see more deforestation and droughts where i live, and unfortunately the AI stuff here (fucking microsoft) isn't making that better.

 

[x65943]

Tbh what upsets me is everything is called AI now

What used to be called path finding in games is AI, what used to be called pattern recognition in image editing software is AI

Every algorithm is not "AI" or if it is, the word has become nearly meaningless

If we are just talking about LLM then I do agree their use is somewhat limited to being a slightly better way to search reddit and other forums for simple answers

With the caveat it will lie to you a lot, and you need to know that to make it useful

 

[kijetesantakalu042]

Tbh what upsets me is everything is called AI now

What used to be called path finding in games is AI, what used to be called pattern recognition in image editing software is AI

Every algorithm is not "AI" or if it is, the word has become nearly meaningless

If we are just talking about LLM then I do agree their use is somewhat limited to being a slightly better way to search reddit and other forums for simple answers

With the caveat it will lie to you a lot, and you need to know that to make it useful

I said generative AI.

I don't think we need to use 400 megawatts of compute for "a better way to search reddit", we could just make search engines suck less?

And if it lies to you a lot, and you need to know the answers already to use it, whats the point?

 

[x65943]

I said generative AI.

I don't think we need to use 400 megawatts of compute for "a better way to search reddit", we could just make search engines suck less?

And if it lies to you a lot, and you need to know the answers already to use it, whats the point?

The issue isn't having LLM as a tool

The issue is shoe horning them into literally everything

If you had to make a conscious choice to use them, then their requirements would be way down

But lately my car and Google home even updated and now default to LLM - something I had no say in (and ofc every Google search now has Gemini at the top)

 

[kijetesantakalu042]

I once saw a conversation on reddit where a mentally ill and depressed person asked something on reddit and someone copy pasting chatgpt had a conversation with them, without that person knowing they were getting therapy from ChatGPT.

I feel like that shows how LLMs existing 'as a tool' has made society worse.

 

[Cranesbill]

For once I agree with Crane here.

TWO things, we agree on now. The Scientology raids, and this

We're making so much progress now

 

[x65943]

I once saw a conversation on reddit where a mentally ill and depressed person asked something on reddit and someone copy pasting chatgpt had a conversation with them, without that person knowing they were getting therapy from ChatGPT.

I feel like that shows how LLMs existing 'as a tool' has made society worse.

one of my brothers is always arguing with me in a groupchat

I have noticed he has become a lot more reasonable since he started using chatgpt to argue with me

on the one hand it's kind of annoying to argue with a bot, on the other hand - it at least doesn't believe in conspiracies

 

[kijetesantakalu042]

TWO things, we agree on now. The Scientology raids, and this

We're making so much progress now

Next you'll even agree with me that the correct way to put the toilet paper on the roll is with the flap coming towards you.

 

[Cranesbill]

Next you'll even agree with me that the correct way to put the toilet paper on the roll is with the flap coming towards you.

OH MY GOSH, ME TOO

 

[AncientBoi]