Early Work on German Histories and Advancements in Technology

2min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

In 2007, I and my friend John Maker started working on German histories and exploring machine learning. At that time, machine learning and deep learning were not widely discussed in humanities. However, we were intrigued by the potential of using OCR to convert textual documents into machine-readable formats. We realized that machine learning could solve the challenge of deciphering handwritten texts and adapting to typewriter font variations. In the 2010s, digital history shifted towards visualization techniques like word clouds, but I found them less personally appealing. Overall, our early exploration of machine learning in history was cutting-edge, even though we were not aware of it at the time.

Mark Humphries is a Professor of History at Wilfrid Laurier University, where he has published widely on various aspects of Canadian history. We invited Mark to do an episode after he reached out to tell us that hearing about how Nathan fine-tuned GPT-3.5 on GPT-4 reasoning helped him get over some hurdles in his own archival research. In addition, he chats about all the things that he'd tried that hadn't worked, and in the process proved himself to be one of the world's leading adopters of AI technology in the field of history.

If you need an ERP platform, check out our sponsor NetSuite: http://netsuite.com/cognitive.

SPONSORS: NetSuite | Omneky

NetSuite has 25 years of providing financial software for all your business needs. More than 36,000 businesses have already upgraded to NetSuite by Oracle, gaining visibility and control over their financials, inventory, HR, eCommerce, and more. If you're looking for an ERP platform ✅ head to NetSuite: http://netsuite.com/cognitive and download your own customized KPI checklist.

Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off.

TIMESTAMPS:

(00:00:00) - Episode Preview

(00:03:00) - The state of AI in history before recent advances like GPT-3

(00:06:00) - Mark's background and how he first got interested in AI

(00:09:00) - Using AI to process and search through vast amounts of archival material

(00:12:00) - The challenges of digitizing and searching handwritten historical documents

(00:15:00) - The massive scale of undiscovered archival material that could be processed by AI

(00:18:00) - How AI can help historians process more archival material than ever before

(00:20:11) - Sponsors: Netsuite | Omneky

(00:21:00) - Limitations on using commercial APIs for private archival documents

(00:24:00) - The cost of processing archival documents compared to hiring human transcribers

(00:27:00) - Comparing GPT's ability to transcribe historical handwriting vs more recent scripts

(00:33:00) - Tracing individuals through complex fur trade records using AI

(00:36:00) - Mark's approach to fine-tuning models for specific historical tasks

(00:39:00) - Challenges teaching AI models the nuances of historical writing

(00:42:00) - Privacy issues limiting the use of commercial APIs on restricted archival material

(00:45:00) - The gap between people's expectations of AI and what it can really do right now

(00:51:00) - Adapting assignments and student expectations to AI's current capabilities

(00:52:28) - Using chain of thought prompting to teach models precise historical tasks

(00:52:46) - Getting high accuracy on keyword tagging of archival documents using fine-tuning

(00:57:00) - Mark has crossed the threshold from promise to accelerating archival research

(01:00:00) - How to frame classification tasks to get the most value from AI models

(01:03:00) - Teaching AI models to pronounce tricky historical names and terms

(01:06:19) - Mark's experience using AI in the classroom over two semesters

(01:09:00) - Assignments that are better suited to the current strengths and weaknesses of AI

(01:11:45) - AI progress in education is lagging due to lack of institutional adoption

(01:12:00) - The trajectory of AI capabilities and the need for humans to exceed the model baseline

(01:27:49) - Lessons from history on economic transitions and the social contract

This show is produced by Turpentine: a network of podcasts, newsletters, and more, covering technology, business, and culture — all from the perspective of industry insiders and experts. We’re launching new shows every week, and we’re looking for industry-leading sponsors — if you think that might be you and your company, email us at erik@turpentine.co.