Nicely done. The illustration is the explanation, and vice versa.
A tell-tale giveaway for me is the two sentence phrasing "This is not X. This is Y." It is typically is found following the summary section as it transitions into the em-dash infused punchy conclusion.
How so many AIs developed this language tick during their training I don't understand. Humans do not write like that. From where does it originate? I can't see how it would be sufficiently present in natural texts. Perhaps it is an effect of indirect self-reinforcement. AI slop infuses out into the web then cycles back into the next training iteration.
Or as you hint, it may be a consequence of user engagement reinforcement learning. The clicks & scrolls feedback may give AI its distinctively off-putting voice.
This is not definitive. This is informed speculation.
Nicely done. The illustration is the explanation, and vice versa.
A tell-tale giveaway for me is the two sentence phrasing "This is not X. This is Y." It is typically is found following the summary section as it transitions into the em-dash infused punchy conclusion.
How so many AIs developed this language tick during their training I don't understand. Humans do not write like that. From where does it originate? I can't see how it would be sufficiently present in natural texts. Perhaps it is an effect of indirect self-reinforcement. AI slop infuses out into the web then cycles back into the next training iteration.
Or as you hint, it may be a consequence of user engagement reinforcement learning. The clicks & scrolls feedback may give AI its distinctively off-putting voice.
This is not definitive. This is informed speculation.
Ah! You see what I did there.