>>140460Hello,
I am unsure whether you are trolling, but real AI is referred to as AGI.
What we know as "AI" is really a "neural network" - so named because each element in the training data is connected to the value of each other element.
It is not "alive" and it does not "think" - it's fully deterministic, if you use the same prompt and seed in Stable Diffusion you will get the same result every single time.
The fact that anyone believes that we have "AI" in the present day is the result of the most successful marketing scam in human history.
Nevertheless, if we did have a neural network that could retrain itself based on user input, or its own outputs, or the outputs of other neural networks, that WOULD classify as AGI and something that could be considered alive by optimists.
So how long until we have that?
Well, the unfortunate truth that I don't like to tell AGI hopefuls is that for neural networks of an even smallish size, training them is a process that is measured in "GPU days" even for ridiculous supercomputer-tier GPUs developed by NVIDIA specifically for this purpose.
So, it's utterly inconceivable that we'll have a neural network that will self-retrain in real time in the near future.
AGI will not happen until 2090.