Geiger , Ph.D., is assistant professor in the Department of Communication, Halicioglu Data Science Institute and Institute for Practical Ethics at UC San Diego. He lives in City Heights. I teach several courses at UC San Diego in communication and data science about the use and abuse of data, digital platforms, algorithms, personalization and artificial intelligence. I was preparing for my winter Data and Culture class when ChatGPT exploded on the scene. I'd used Natural Language Processing methods before GPT-1, but ChatGPT felt different. When I gave it my favorite assignments, it always passed my rubrics with remarkably average essays. Due to how ChatGPT works, I knew it would be difficult or impossible to definitively detect its use. On a whim, I had ChatGPT generate a syllabus for my class. It gave relevant topics and readings, including several new readings from my favorite scholars — which did not exist. I found similar “hallucinations” when I asked for direct quotes from a reading. I began to immunize my assignments, crafting prompts that ChatGPT would get wrong. But I soon realized that this was the wrong direction. One reason is because these systems are rapidly developing, so today's flaws may be fixed tomorrow. Commentary Opinion: Advancements in artificial intelligence pose questions about the future of humanity June 16, 2023 More importantly, I realized there were striking parallels to Wikipedia, the online encyclopedia anyone can edit, which I've edited since 2004 and extensively researched. In the mid-2000s, there was panic about students using this new, uncertain and unreliable information resource. Schools banned Wikipedia while school officials shamed students who read it, and some even argued Wikipedia undermined the idea of truth. Yet the people who wrote Wikipedia didn't naively trust it. Wikipedians agree it should never be cited definitively and could be dangerous if used uncritically for high-risk decisions. Instead, it was a resource, a starting point, an inherently imperfect and incomplete guide to reliable sources. These very imperfections mean you can observe how knowledge is represented and contested on Wikipedia. You can see the revision history of every article and spend […]
Click here to view original page at Opinion: ChatGPT is this generation's Wikipedia. We have an opportunity to learn from the past.
© 2023, wcadmin. ©2023. All rights reserved, Writers Critique, LLC Unless otherwise noted, all posts remain copyright of their respective authors.