I must be among an extremely small group of people (n=1?) that have both 1) trained a frontier LLM and 2) designed and synthesized custom viruses in a lab with my own two hands.
And I think that the takes on AI killing us all by creating dangerous viruses is total bogus.
@mehrantsi@evnsnclr I am not able to understand.. how simply network without weight can turn the flight simulator. These nerves are like room wiring which to do not have trasistor or register to run the home appliances.
Until these do not have weight can not do any task without policy train
@kutsch_d@evnsnclr@OpenAI So next extension would be to map weights, capture every signal like the projects where we mapped the brain waves by showing the person perticular image or tell them to read and see what circuits are being fired and train synthetic NN to predict just by monitoring brain waves..
@evnsnclr@OpenAI The main part is , these brains are just a connections it do not have weights.. and the wieghts are even more important to drive any perticular skill which requires weight ( like nobs to fine tune the skills).
@evnsnclr@OpenAI I tried similar things by using fruitfly brain to control so101 robot arm in mujoco..month ago
But it performed way poor than normal even random NN.
@moshhamedani I agree I built enterprise nms zero line of code i have written but yes it took months not days.. but I suppressed 4 years of development to 3 month and 300 usd claude ..
@ProfBuehlerMIT Why i say this.. our body is collective functions of different agents which runs overall function of a body.. and thus making ourself (an individual and its internal function working together as society) concious
@drummatick Any command you run is first prechecked by lower model sonnet. Only in claude .. it takes the decision and then passes to main.. thus your command burns tokens