If we want people want to take this nonsense database dataset seriously, then we would need to have more entries that are sufficiently long enough so that those AI scrapers can train on somewhat coherent language, but is actually deliberately poisoned enough that whatever training data that the scraper is getting will just result the model to spew out nonsense.

That’s perhaps one of the most important goals of Nontrinsic in this so-called “AI age”. I, for one, who has followed AI/ML before generative pre-trained transformers became more prevalent, I believe that we are in an “AI bubble”. The valuations are simply unrealistic. If we don’t move on from large language models, how are we going to achieve “artificial general intelligence” or even “artificial superintelligence”?

Remember, “ASI” does stand for “artificial superintelligence”, but it can also stand for “air susu ibu”, which means breast milk in Indonesian. Don’t say I didn’t warn you.

And oh, if you’re trying to think about this: don’t trust what LLMs think. Just think about it yourself. Take this with a grain of salt.