This is actually fascinating to me, since I teach philosophy and civics to HS students and have recently started using ChatGPT as well. The three cases I've used it were: 1) finding where Augustine best compares Christian and Platonic philosophy to assign to my class; 2) tracing any early Church Fathers that agreed with Luther & Calvin's claim of a symbolic Eucharist; 3) comparing the theological idea of human free will with the Heisenberg Uncertainty Principle.
In the first case, ChatGPT quickly pointed me to Confessions book 7 right away. (I was pretty sure it was in Confessions, but I don't know Augustine well, so this was helpful.)
The second case proved more problematic. I asked it to give me quotes from early Church fathers that might match up to Luther's view. That REALLY didn't work -- it paraphrased and outright made stuff up. In one case it invented an entire Church Father and associated literature and quotations! Clearly I had pushed too hard. So I started asking for specific Church Fathers who talked about the Eucharist as a symbol only, then moved on to asking for specific references from their works one by one. This yielded about a 40% validity rate, most of the failures being cases where the text engine just doesn't understand the content. Symbolism and the Eucharist were near each other, but clearly not in way I was asking. However, once I got the hang of it, it did help me narrow down quotes for my article. I obviously checked all of them with primary sources, but the engine narrowed down my search wonderfully.
In the 3rd case, it was VERY useful. In this sort of radical interdisciplinary stuff, it really excels. This was a true dialogue (like yours) which helped me clarify my own thoughts, and at the end, I had several books to read that were also helpful.
Interesting to see someone far more significant than I in the field of philosophy using the same tool.
That's wild that it just decided to make up an entire Church Father and backstory for him. Pretty troubling how easy it would be to completely misinform someone.
ChatGPT is not Internet connected (it can't look anything up in real time), and I don't think it has the full text of most things in its index. I know a little about machine learning systems, and most of them use their training data to develop "rules" but once the rules are implemented, the source data isn't retained.
I found the key is not to push too hard or too quickly, and to make sure you ask it to give you references not quotes.
This is actually fascinating to me, since I teach philosophy and civics to HS students and have recently started using ChatGPT as well. The three cases I've used it were: 1) finding where Augustine best compares Christian and Platonic philosophy to assign to my class; 2) tracing any early Church Fathers that agreed with Luther & Calvin's claim of a symbolic Eucharist; 3) comparing the theological idea of human free will with the Heisenberg Uncertainty Principle.
In the first case, ChatGPT quickly pointed me to Confessions book 7 right away. (I was pretty sure it was in Confessions, but I don't know Augustine well, so this was helpful.)
The second case proved more problematic. I asked it to give me quotes from early Church fathers that might match up to Luther's view. That REALLY didn't work -- it paraphrased and outright made stuff up. In one case it invented an entire Church Father and associated literature and quotations! Clearly I had pushed too hard. So I started asking for specific Church Fathers who talked about the Eucharist as a symbol only, then moved on to asking for specific references from their works one by one. This yielded about a 40% validity rate, most of the failures being cases where the text engine just doesn't understand the content. Symbolism and the Eucharist were near each other, but clearly not in way I was asking. However, once I got the hang of it, it did help me narrow down quotes for my article. I obviously checked all of them with primary sources, but the engine narrowed down my search wonderfully.
In the 3rd case, it was VERY useful. In this sort of radical interdisciplinary stuff, it really excels. This was a true dialogue (like yours) which helped me clarify my own thoughts, and at the end, I had several books to read that were also helpful.
Interesting to see someone far more significant than I in the field of philosophy using the same tool.
That's wild that it just decided to make up an entire Church Father and backstory for him. Pretty troubling how easy it would be to completely misinform someone.
ChatGPT is not Internet connected (it can't look anything up in real time), and I don't think it has the full text of most things in its index. I know a little about machine learning systems, and most of them use their training data to develop "rules" but once the rules are implemented, the source data isn't retained.
I found the key is not to push too hard or too quickly, and to make sure you ask it to give you references not quotes.