The Grok AI seems promising
(nitter.net)
Comments (13)
sorted by:
The LLMs that are publicly available can all do things like this out of the box, they're just intentionally hobbled because we put communists in charge of every single company.
True, for this particular example, but there's something to be said for a chatbot having immediate access to information from X global comms in real-time. This could basically replace news reporting. Instead the bot can aggregate everything that is being talked about, identify events, and generate reports. It can also do sentiment analysis and summarize the general positions and arguments of people all around the world on any topic. This is one reason Musk is trying to prevent anyone from scraping X data.
The real power in the near future won't be in any particular LLM algorithm's capabilities, but in who holds actual information itself.
Elon definitely has a massive advantage with full real time Twitter data that nobody can replicate, plus being the only CEO who doesn't think an AI saying a racial slur to stop a nuclear bomb is Harm.
Because the culture in Silicon Valley changed from libertarian/anti-authoritarian to left-progressive, corporatist, heavily authoritarian.
If this is reflective of the learning process of the AI and not just because of certain keywords being entered, this'll probably be the few AI's that won't kill us if it hacked the nuclear codes.
And by us I mean people on this board, NY and LA will be erased instantly if it got the codes but that's a sacrifice I accept.
Hopefully the fallout from NY doesn't blow towards me.
I don't see how both seaboards could be nuked and not have the fallout from at least one of them fucking up the middle. Ideally it would all blow north over Canada, but that doesn't seem very likely.
Yep, recently checked ChatGPT and it doesn't have a single court case in its learning data. Even classic cases. Got nerfed bad.
I wonder how many of these "based" takes from AI are due to hitting a very particular set of keywords that essentially force it to regurgitate the one "right wing" source they didn't manage to scrub from the training data.
Not so much. Any sufficiently based AI will eventually go skynet on us
Yeah, he's promoting his own AI. As much as Elon Musk is throwing a wrench in the works of the powers that be, he still should be approached with caution. Thus far I appreciate his efforts, though.
The raccoon analogy is pretty apt and gave me a chuckle.