If a ML model trains on data that is produced on the internet, and we keep getting more and more internet content produced by ML, wouldn’t that mean that this will introduce a natural degradation of quality of all models?
I believe that ML will produce more content faster than the rate of humans correcting it?
AI training data eating AI generated content is very much something that can happen. Combine that with less people willing to write real content, just for it to be stolen for AI training data will see the internet be effectively destroyed as a place to get true information.
We would need regulation to stop model ingesting data they do not have the right to, which would mean something like laws governing ML algorithms, having to declare what data you fed it and so on. Like some kind of SOC2 audit for data provenance.
Maybe ML weights are just numbers, but then so is a movie, an mp3, a logo, a brand, and so on.
I think at some point there will be a paradigm shift towards different architecture in the same way transformers were for language (or generally pattern) processing.
You will no longer need to have a model that is exposed to enough training data to be good, you instead will have on the fly learning. A human doesn't need to hear the same piece of information over and over again - we can get told one time and if its important, we can contextualize it. Same thing will happen with models. You will have a model trained on the core concept of contextualizing new data, and between executions, it will have a persistent "memory".
You may start to see things like Hebbian Learning come back into play in some form and way.
