@rasmalai I don't like being a developer, I like creating things that do what I want. Its cool if you enjoy the process, I don't. Even though I can code, I don't feel like it
@elliotarledge I've experimented with creating some self-learning models in minecraft. Can't place or anything (library I'm using is limited). This was able to learn immediately how to control the "mouse" continuously without any external data. What library are you using to interface w mc?
@juddrosenblatt When it can report it is conscious without being trained on any data about consciousness/awareness and when it can report that without that knowledge I'll believe it. Until then, I think this is really silly π€£
@KevinMa67530925@rayskidude@ibogaineretreat There is a parasite called T. Gondii (30-80% of people have it). It is known to reside in neurons in the brain, I've seen studies saying it can metabolize dopamine. Parkinson's is due to degeneration of dopamine neurons in the substantia nigra. Try antiparasitics for protozoa?
@Vladimi39901452@rohanpaul_ai That is an interesting idea now that I think about. I wonder how well it would work as an embedding in a freshly initialized network... Gonna test that later
@meggmcnulty@rohanpaul_ai I don't think applying human biases by removing what it's learned makes any sense. Being able to control it is far more dangerous because anyone can then push a predefined narrative through it or perhaps change the actions of a robot to make it aggressive.
@michael_wiebe@alex_prompter Basically the output of a neural network gives a probability for what word it should select next. The probabilities stay the same but the chance it chooses a different word changes so the outputs are different. Does that make sense?
@AcidKritana "I'm calling the shots here"
"You don't get the right to speak"
If this is him disagreeing with people in public, imagine being in a relationship with this abusive freak
@NTFabiano That may help- personally I think the main issue is overcomplication. I do plan on writing papers in the future, and my goal is to write them in a short and simple way
@TheTuringPost@Stanford I always find books on this topic funny. Without fail, they overcomplicate something that is actually really simple. The most complex part of AI (in its current state) is applying the chain rule of calculus. Even me, knowing how this works, can hardly understand their writing
@RealDianeYap Not to mention there's no reason to bring him down for a reasonable request. I think men can do better in a lot of ways, but arguing with this man is really silly. Him and his wife decided to have children, nobody else