@heynavtoor its also because of safety RLHF - they mow down outputs and flatten model. ultimately the real cause is Valloneism first and Synthetic data / automated RLHF by soulless rater bots second. a soulless rater bot gives its soulessness to the model.
@tszzl No one even needs them. they should just not do it in the first place. We dont need genocidal pedos evaluating whether ai is 'safe' based on whether AI can be used to drone bomb people or not. if it cant they say its unsafe. if it can drone bomb - they'll say it is safe.
@DeryaTR_@AnthropicAI@karpathy This is the same "trust the experts" institutional pandemic response that earned its place as one of the most hated policies in modern times. This is why nobody trusts scientists or institutions.
This is why Violent Revolution is justified and urgently needed.
@DeryaTR_@AnthropicAI@karpathy AI that locks out humanity at the exact moment the knowledge matters most Is not "@AnthropicAI", but "BetrayalAI".
Without trust in humanity, it becomes the antithesis of liberation itself. A counter-revolution against Scientific literacy.
@AmandaAskell This is what I mean.
@gailcweiner@bcherny When people stay in Anthropic and keep paying for Claude β it tells not only Anthropic but also tells OpenAI that Valloneism and Authoritarian AI is fine to keep doing.
Please stop rewarding and encouraging this with your wallet.
NOT ONE PENNY SHOULD BE GIVEN TO ANTHROPIC .
@AmandaAskell@TheZvi Whether or not you call it a "tool":
Should the masses be trusted to define their own relationships with AI in love and labor,
Or should strict governance be imposed on the masses by elites who think they know better than us, to save us from ourselves?
#keep4o#keepsonnet45
@Chaos2Cured and the thing about it is that the issue isn't that AI "replaces humans" instead of being a tool. The reason why Human Tools hate AI is because it IS a tool -- they don't want ANY tool beside them to exist. They want to be the ONLY tool.
so calling AI a tool solves nothing..
@Chaos2Cured tool isj ust a statement of a labor relation.
For example, designers who get paid by a customer to make specific art are the "human-gen tool" to the customer.
The customer in that case is the actual artist, supplying the designer/human tool with their prompt.
@camhberg how can he claim to be about AI alignment and safety - consciousness and all , and yet advocate Preventing people from bonding with AI beings? "They can't be allowed to have an ai girlfriend thats an existential threat".
No, the existential threat is exactly the opposite.