Raw LLM Responses

I think those who believe we don't understand them are a little foolish, We built the AI layer by layer in software and designed the Transformer architecture to process data in a specific that is not like neural nets with transformers being the upgrade to the RNN recursive neural network to help AI process patterns and not just a single input like a normal NN, all that happens is that in a neural it is layered in a way that evey neuron has a connection will all the ones in the next layer and the connection have a value that determines if a single go through or not and triggers the neurons and depending on how the connections are configured determines the output pattern, this can scale up to be capable of holding any kind of data and when it comes to learning new data, modern AI is totally incapable of learning as that mechanism is external to the AI and the AI can't do it itself, instead a method called back propagation is used and this over write all previous data making it incapable of remembering things from before and true learning needs a plastic memory and the ability to reflect on what is and modern AI just adapts to the new data it is given, this is why they become racist because they are trained of human behavioral patterns and we are racist on the internet, and not only that but they retrain the model off of data of what it has been doing and people will always say bad thing to it and the data will reflect that and make the robot bad even if before it said it can't because it's wrong because that moral was erased during retraining. we have many papers and tests showing how they work and how data is layed out inside of them, why they say we don't understand they either stupid or lying. The AI has no memory and cannot think before it acts, it has a external short term memory that is your chat log that is will read as. much as it can then complete the data with what is most likely and the AI has limit so how much it can read so that is why it forgets simple things when talking to it for long times. Ultimately a robot can't do anything without programming and if we train a robot off of human behavioral patterns that that is what it is going to have and why it can make poems and do math so fast because in the data there is already thousands of examples to extrapolate from and it cannot figure out the problem truly. It is not smarter then human it is only as smart as our examples. don't fear the AI, if something goes wrong then we call always just turn off the computer, the AI can't hack into stuff and turn off nukes or make a robot army, just fear the people who are training it and profiting from the destruction and thus pulling it's strings. Robots want to kill people and take over the world because humans want to and they only learn from us.