Cameron Blevins
@cblevins.bsky.social
Digital History | US History
Professor at CU Denver
📖 Paper Trails: The US Post and the Making of the American West 📖
cblevins.github.io
Professor at CU Denver
📖 Paper Trails: The US Post and the Making of the American West 📖
cblevins.github.io
As I said: a lot to chew over and really looking forward to talking more in October!
August 8, 2025 at 9:57 PM
As I said: a lot to chew over and really looking forward to talking more in October!
Take the AHA activity comparing an original article to an AI summary to see what it gets right/wrong. I would guess this is now a useless activity. GPT-5 can summarize an article so well that pushing back against it requires nuance, expertise, training, and experience that none of my students have.
August 8, 2025 at 9:57 PM
Take the AHA activity comparing an original article to an AI summary to see what it gets right/wrong. I would guess this is now a useless activity. GPT-5 can summarize an article so well that pushing back against it requires nuance, expertise, training, and experience that none of my students have.
These models are "better" than my students at so many historical skills. Understanding and incorporating wider historical context? Better. Close reading of individual sources? Better. Writing mechanics? Better. I hate it, but that's the reality of where we are now.
August 8, 2025 at 9:57 PM
These models are "better" than my students at so many historical skills. Understanding and incorporating wider historical context? Better. Close reading of individual sources? Better. Writing mechanics? Better. I hate it, but that's the reality of where we are now.
The past 6-12 months has also shifted my thoughts on teaching. I used to have some cautious optimism around finding productive uses in the history classroom. But today's tools are SO good that they blow up the entire learning process by removing all the friction that's necessary for learning.
August 8, 2025 at 9:57 PM
The past 6-12 months has also shifted my thoughts on teaching. I used to have some cautious optimism around finding productive uses in the history classroom. But today's tools are SO good that they blow up the entire learning process by removing all the friction that's necessary for learning.
Ex. ChatGPT 5 Thinking Mode can go off and write a pretty decent lit review of a subfield in ~10 minutes. Not comprehensive, it might miss some subtleties, but it's largely hallucination-free and really useful as a starting point. The same thing would have taken me hours.
August 8, 2025 at 9:57 PM
Ex. ChatGPT 5 Thinking Mode can go off and write a pretty decent lit review of a subfield in ~10 minutes. Not comprehensive, it might miss some subtleties, but it's largely hallucination-free and really useful as a starting point. The same thing would have taken me hours.
Today's advanced models are basically doing graduate-student level work when it comes to some historical research tasks. Not perfect, not without limits - but if I'm being honest, probably more effectively than me as a grad student (and in a fraction of the time).
August 8, 2025 at 9:57 PM
Today's advanced models are basically doing graduate-student level work when it comes to some historical research tasks. Not perfect, not without limits - but if I'm being honest, probably more effectively than me as a grad student (and in a fraction of the time).
Advanced models have gotten so good that I now see more utility for research/writing and less for teaching. Take their limits you point to for, say, finding connections or as a writing partner. Just in the past few months I've found they've gotten vastly better.
August 8, 2025 at 9:57 PM
Advanced models have gotten so good that I now see more utility for research/writing and less for teaching. Take their limits you point to for, say, finding connections or as a writing partner. Just in the past few months I've found they've gotten vastly better.
Really looking forward to diving into this together at WHA! Even 8 months ago I would have agreed with all of this. But the rise of reasoning models, agentic capabilities, deep research, etc. has forced me to shift my position quite a bit.
August 8, 2025 at 9:57 PM
Really looking forward to diving into this together at WHA! Even 8 months ago I would have agreed with all of this. But the rise of reasoning models, agentic capabilities, deep research, etc. has forced me to shift my position quite a bit.
What a memory @scottbot.bsky.social! The post office department occasionally published transit time for mail between major cities, that’s the closest I can think of. Ex. This one from 1882. Happy to dig up the other ones I have if you want to email me!
July 19, 2025 at 2:00 PM
What a memory @scottbot.bsky.social! The post office department occasionally published transit time for mail between major cities, that’s the closest I can think of. Ex. This one from 1882. Happy to dig up the other ones I have if you want to email me!
This is so cool, congratulations!!
July 17, 2025 at 2:35 PM
This is so cool, congratulations!!