253
this post was submitted on 06 Dec 2023
253 points (95.7% liked)
Technology
59197 readers
3468 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
By answering questions? We are general intelligences that can answer questions. Oh shit oh fuck what am I doing talking.
Hey guess what, we general intelligences are capable of terrible things.
That was exactly my point. AI isn't, though. All it can really do is respond.
I'm sure the military is so excited about AI because of its ability to "respond".
If you can get chatgpt to drive your murder drone I'd be very impressed. Telsa can't figure it out in 2d.
My brother in Christ, the AI revolution is more than chatgpt.
Call when it gets dangerous enough to be an issue.
Im bot sure if you are aware but thats litterally what makes ai so usefull its just responding to external inputs and doesent habe to be programmed value for value because it getsTrained with datasets and chat GPT isn't gonna hurt a fly the reason its m Rated is because the idiots who made it didn't filter the input content whilst web scraping its litterally too stupid to Funktionass a weapon except for misinformation which it outputs regardless oft its age rating openai is just a buncha cucs who switched to a close source system and can't actually make any gold company decisions
TL;DR: Fuck openai
FWIW I work in the field and agree with this. LLMs in the current state are not so dangerous they can't be released to public. Generative image and video models are a much bigger threat, but that was largely something which came from open source.
If we really want to pearl clutch, it is NVIDIA which is really propping open this Pandora's box in terms of putting the capability in irresponsible hands
Are the cards really powerful enough for so much fuss?
Definitely. An A100 system is around $10k which is expensive, but definitely in reach, and you need two of them to run a 70B parameter model. Possibly one if you are clever about it.
And you can still do plenty of damage with a $1000 consumer grade GPU. Most deepfake videos are trained on these platforms.
Okay, so let's do a thought experiment, and take off all the safeguards.
Oops, you made:
Saying "don't misuse it" isn't enough to stop people misusing it
And that's just with chatgpt - AI isn't just a question and answer machine - I suggest you read about "the paperclip maximiser" as a very good example of how misalignment of general purpose AI can go horribly wrong
I was going to say that a well-determined individual would find this information regardless. But the difference here is that it being so easily accessible would increase the risks of someone doing something reaaaally stupid by a factor of 100. Yikes.
None of that is complicated.
For you or many others, for sure it won't be complicated. The world is vast, and the environment you are in is very specific to you. Many other kids may have phones, sure, but they are not in the same environment as you or me.
Some non-sciency kid will have a hard time getting to do what their edgy mind wants them to do, unless an AI guides them mini-step by mini-step.
I don't think AI, especially chat bots, will be more useful than a youtuber. It's not particularly easy to make powerful explosives, and gun powder is kind of trash for bombs. I'd imagine chatgpt would blow up more curious kids than aid them lol
The last 2 already exist it's called stable diffusion. And for awhile Bing did it too.
I mean half that is deviant art and you can look up how to make explosives on youtube chem channels or in books. It's not hard to rig up a custom detonator if you can get the energetics.
What are we going to do with these Universities, just teaching anyone how to be a chemical engineer!!?
Jesus some of you guys completely miss the point.
ChatGPT was very far from the first publically available generative AI. It didn't even do images at first.
Also, there are plenty of YouTube channels which show you how to make all sorts of extremely dangerous explosives already.
But the concern isn't which was the first generative ai - their "idea" was that AIs - of all types, including generalised - should just be released as-is, with no further safeguards.
That doesn't consider that OpenAI doesn't only develop text generation AIs. Generalised AI can do horrifying things, even just by accidental misconfiguration (see the paperclip optimiser example).
But even a GANN like chatGPT can be coerced to generate non-text data with the right prompting.
Even in that example, one can't just dig up those sorts of videos without, at minimum, leaving a trail. But an unresticted pretrained model can be distributed and run locally, and used without trace to generate any content whatsoever that it's capable of generating.
And with a generalised AI, the only constraint to the prompt "kill everybody except me" becomes available compute.