An interesting application of AI to Lucretius and Epicurean philosophy.
Very interesting! Only had time for a quick scan but I have to comment on this threshold point:
Salutations. As Lucretius, the ancient Roman poet and philosopher, my magnum opus "De Rerum Natura" (On the Nature of Things) serves as a grand exposition of Epicurean philosophy. This didactic poem, written in the first century BCE, is a testament to my deep admiration for the teachings of Epicurus, the Greek philosopher whose atomistic and materialistic view of the world profoundly influenced my thinking.
The poem, spanning six books, is a comprehensive attempt to explain the nature of the universe, the role of humanity within it, and the principles of Epicurean thought. Its central thesis is that the universe operates according to natural laws, without the intervention of divine beings. It emphasizes the importance of seeking tranquility and freedom from fear, particularly the fear of death and the gods, a state known as "ataraxia."
To me that is already interesting how the AI system may have been fed Lucretius' poem, but I certainly don't think it got that underlined sentence from the poem itself, but from outside commentary. Had it reasoned primarily from the poem, I would expect it would have picked up "pleasure" as what is emphasized, rather than tranquility or freedom from fear, which might have been picked up as implications but not as the first thing one would mention after summarizing the poem for oneself.
The thing about the language models is that they're just a math equations with assigned linguistic assignments. So as they go along parcing out they're sentences the math side is looking for the most likely continuation of the sentence or paragraph. So what ever the model was trained on led it to believe that that was the most likely sequence of words. It was likely trained on a whole collection of philosophical works aswell as "the Pile". I had at one point considered doing the same thing, training an epicurean chat bot and seeing what it would output. But honestly I'm really disappointed with the reliability of the data coming out of the current models. From what I've seen it will be another 3-4 full evolutions of the tech before it's really reliable. Right now it's more like a parlor trick than a real tool.
The thing about the language models is that they're just a math equations with assigned linguistic assignments. So as they go along parcing out they're sentences the math side is looking for the most likely continuation of the sentence or paragraph. So what ever the model was trained on led it to believe that that was the most likely sequence of words. It was likely trained on a whole collection of philosophical works aswell as "the Pile". I had at one point considered doing the same thing, training an epicurean chat bot and seeing what it would output. But honestly I'm really disappointed with the reliability of the data coming out of the current models. From what I've seen it will be another 3-4 full evolutions of the tech before it's really reliable. Right now it's more like a parlor trick than a real tool.
I found the same thing. I thought it could be an effective research assistant, but it is unreliable, never up-to-date with the latest research, and it presents a huge opportunity to exploit confirmation bias by training it to answer selectively, so I am not impressed by ChatGPT.
When I have tried asking AI about Lucretius and ask for references it typically says it cannot access the Latin text and then I argue with it about copyright. When it does give a reference it is usually incorrect. Success only occurred once or twice and then I gave up. However, it is proving useful as a dictionary.
