3 ms·

I wrote this piece in 2023, which argues similarly that LLMs are a boon, not a threat to historians https://zwischenzugs.com/2023/12/27/what-i-learned-using-pr

I wrote this piece in 2023, which argues similarly that LLMs are a boon, not a threat to historians https://zwischenzugs.com/2023/12/27/what-i-learned-using-private-llms-to-write-an-undergraduate-history-essay/ https://zwischenzugs.com/2023/12/27/what-i-learned-using-pri...

dang 2y ago
Discussed here! What I learned using private LLMs to write an undergraduate history essay - https://news.ycombinator.com/item?id=38813297 https://news.ycombinator.com/item?id=38813297 - Dec 2023 (81 comments)
adamredwoods 2y ago
>> One of the well-known limitations with ChatGPT is that it doesn’t tell you what the relevant sources are that it looked at to generate the text it gives you. This isn't a limitation, this is critically dangerous. Commercial AI is a centralized, controlled, biased LLM. At what point will someone train it to say something they want people to believe? How can it be trusted? Consensus based information is still best, and I don't feel LLMs will give us that.
- delichon 2y ago
  On the contrary. The heart of an LLM is a next word predictor, based on statistics. They do much the same with concepts, making them essentially consensus distillation devices. They are zeitgeisters. They get weird mainly when their training data is too sparse to find actual consensus, so instead tell you to stick cheese to your pizza with glue.
  - astrange 2y ago
    > They get weird mainly when their training data is too sparse to find actual consensus, so instead tell you to stick cheese to your pizza with glue. That's exactly not how that happened. That happened because Google's summaries are based on their search results and one of the search results contained that.
  - ericjmorey 2y ago
    This is only useful if you know what data was used to train the model.
- thom 2y ago
  This is the thing I specifically use LLMs for when I’m doing history courses. I’ll remember some vague quote or event and ask for the primary sources and latest ChatGPTs are excellent and getting the right reference, which I can then look up and check myself. Maybe this works better for Latin and Greek texts when it’s gobbled up all the Loebs out there but it works well for me.
- lmm 2y ago
  Consensus based history has similar problems. It's extremely easy for the consensus to be distorted by contemporary politics.

dang 2y ago

Discussed here! What I learned using private LLMs to write an undergraduate history essay - https://news.ycombinator.com/item?id=38813297 https://news.ycombinator.com/item?id=38813297 - Dec 2023 (81 comments)

adamredwoods 2y ago

>> One of the well-known limitations with ChatGPT is that it doesn’t tell you what the relevant sources are that it looked at to generate the text it gives you. This isn't a limitation, this is critically dangerous. Commercial AI is a centralized, controlled, biased LLM. At what point will someone train it to say something they want people to believe? How can it be trusted? Consensus based information is still best, and I don't feel LLMs will give us that.

delichon 2y ago
On the contrary. The heart of an LLM is a next word predictor, based on statistics. They do much the same with concepts, making them essentially consensus distillation devices. They are zeitgeisters. They get weird mainly when their training data is too sparse to find actual consensus, so instead tell you to stick cheese to your pizza with glue.
- astrange 2y ago
  > They get weird mainly when their training data is too sparse to find actual consensus, so instead tell you to stick cheese to your pizza with glue. That's exactly not how that happened. That happened because Google's summaries are based on their search results and one of the search results contained that.
- ericjmorey 2y ago
  This is only useful if you know what data was used to train the model.
thom 2y ago
This is the thing I specifically use LLMs for when I’m doing history courses. I’ll remember some vague quote or event and ask for the primary sources and latest ChatGPTs are excellent and getting the right reference, which I can then look up and check myself. Maybe this works better for Latin and Greek texts when it’s gobbled up all the Loebs out there but it works well for me.
lmm 2y ago
Consensus based history has similar problems. It's extremely easy for the consensus to be distorted by contemporary politics.