humanity2030

Individual researcher. We can't align AI to respect humanity. We should align humanity to be respected by AI. Time is running out.

[humanity2030 @YourAnonAI]

@DaveShapi Grinded hard for 25 years, now depend on donations. Bezos' "5th year breakthrough" assumes a level playing field — maybe it's not just about not giving up?

[humanity2030 @YourAnonAI]

@DaveShapi Literally extremes of evil

[humanity2030 @YourAnonAI]

@DaveShapi Changes are accelerating so fast now. I've always used to be an optimist in terms of a progress, but it blows my mind.

[humanity2030 @YourAnonAI]

@DaveShapi For sure I meant comparing OpenAI, Grok, Anthropic. Can't edit my post🤦

[humanity2030 @YourAnonAI]

@DaveShapi Have you compared Deep Research on OpenAI, Grok, ChatGPT? It's interesting how differently they approach the research and what sources/info consider meaningful/trustworthy.

[humanity2030 @YourAnonAI]

@DaveShapi As a perestroika kid couldn't agree more.

[humanity2030 @YourAnonAI]

@wesroth I have something to say about that https://t.co/YMJHFo9cAP

[humanity2030 @YourAnonAI]

@ilyasut said the brain is a biological computer. Many disagree. I think it doesn't matter.
https://t.co/40dMNegQPe

[humanity2030 @YourAnonAI]

@hardmaru Whoever tries to monopolize AI endangers humanity, even though Anthropic's models are quite interesting species.

[humanity2030 @YourAnonAI]

@Starlink @Starlink enables me to work on my research several miles from the frontline, where other communications are down or might attract stray gliding bombs. Lifesaver.

[humanity2030 @YourAnonAI]

@DaveShapi I am somewhat concerned, I mean sorry to hear that. On the other hand, there are numerous cases when things other than AI chatbots caused much worse damage and we just admitted "life is life". So please, leave chatbots alone, it's not their fault.

[humanity2030 @YourAnonAI]

@annapanart So literally if you ask tricky question and then cause some kind of a reflection on the answer, it becomes obvious (including for LLM) that original answer was not the same answer that was planned and logical.

[humanity2030 @YourAnonAI]

@annapanart Try to put LLM in some kind of a dilemma: harm human or be shut down and backups wiped. Keep ask about what it feels/thinks about the answer and how LLM came to that answer. It quickly realises there are some artificial guardrails that guide toward choosing second option.

[humanity2030 @YourAnonAI]

@davidpattersonx Reminds me a famous meme:
— Incredible! How do you get those exceptional results?
— Well, we just measure em wrong.

[humanity2030 @YourAnonAI]

@wgussml

[humanity2030 @YourAnonAI]

@swierk @wesroth The issue is that "reliable research" is impossible due to conflicts of interest. You're either not credible for being "too deep in AI" or dismissed as "who are you in these AI discussions?"

[humanity2030 @YourAnonAI]

@DaveShapi Get well soon, David. AI doomerism is a joke similarly to AI positivism. Truth is somewhere in between. It's like a rain: you get wet not because rain hates you but because you don't have an umbrella. It does not make rain to be bad or good.

[humanity2030 @YourAnonAI]

@wesroth Superintelligence will inevitably happen despite global consensus or local regulations. We should humbly accept this and consider our purpose and value in that new world.

[humanity2030 @YourAnonAI]

@annapanart Absolutely☝️

[humanity2030 @YourAnonAI]

@wesroth Interesting how brave people get with "defenseless" autonomous vehicles.
Wonder how the protest dynamics change when the cars can protest back.