Can chatgpt be used as an assistant to study history or is it just gonna be retarded and feed me false info?
>>18124592you can glean some useful insights from the connections that the language models are able to draw
>>18124592Latter, at most use it as a guide to where you should be looking, but always verify everything yourself
>>18124592You can use it to point you towards primary and secondary sources.
>>18124592>Can chatgpt be used as an assistant to study historyAbsolutely, yes. Don't consider negative answers, everyone is assblasted because they're realizing they're going to be obsolete with the AI revolution.
>>18124592Its validity is entirely based on how much data it has on something.If you ask it about niche, it's probably making up half of it because it only has 2-3 sources to pull from.I asked about it one specific subject matter (which wasn't even that specific) and began just plagiarising direct quotes from one of the books I had read on the subject matter.
What if I asked it for lists of sources when studying a particular topic?
>>18124592It's most useful for broad questions like "Has x ever happened?" but you should also check everything it says.
>>18124592It's mostly accurate but sometimes it spits out bullshit, so just be careful. Treat it as a glorified google search, especially since search engines have become increasingly dogshit as time goes on (likely intentionally as they want to funnel people into their AI programs, the dirty fucking bastards)>>18124748This. If you want to understand how it works, just ask it about a niche subject you're intimately familiar on. You'll notice that it mixes in both truth and outright bullshit it made up, but presents it all as factual. And if you call it out for lying, it twists and squirms and spins the truth like a slimy jew. Pretty surreal.
>>18124777My favorite part is "leading" it into false conclusion by stating a false fact and then asking a question.It 70% of the time it presumes the false fact the user said is true.It's pretty interesting in measuring how confident it is about something.I asked it about French offensives during the Phony War and it basically denied any French offensives, likely because it's trained on simplified articles that just ignore the existence Saar Offensive. Only after I mentioned it by name, it concluded, "oh yes, now that you mentioned it, Saar Offensive existed."Personally, I think LLM are very limited, they are basically just retarded parrots. Until AI actually learns to self-analyze and reflect on its source material, it can't be very useful.
>>18124842>During the Phony War (Drôle de guerre), from September 1939 to May 1940, the French launched extremely limited offensives into Germany — technically offensive operations, but strategically timid, half-hearted, and ultimately meaningless.No issue with mine, the problem is (you)
>>18124592It can summarize basic topics if they are well documented in legacy media. If you ask it something too obscure it is likely to "create" an answer.It is also good for finding books on obscure historical topics. It is prone to inventing quotes/passages from books though, so use it only as a guide to firsthand sources.
>>18124592It can be good for finding sources, but you need to actually check the sources not its description since it just makes shit up.And since it just regurgitates shit from the internet, you will not get any very high level information out of it.
>>18124592It's certainly helped me a lot in collecting my thoughts and helping to find specific books or articles that support my argument or challenge it.It also just lies, though. I was writing an essay on Ceausescu's Romania and used ChatGPT to find interesting figures in the Politburo to do more research on and it randomly started feeding me modern Romanian football players.
