Summary of “ChatGPT for Academic Research: The Good, the Bad, and the Ugly” by Hannah Rozear (Duke Librarian for Biological Sciences & Global Health)

In case you missed it, Hannah Rozear (Duke Librarian for Biological Sciences & Global Health) gave a talk this week entitled ChatGPT for Academic Research: The Good, the Bad, and the Ugly (watch here). Here’s my summary:

The Good:  ChatGPT has the potential to:

  • accelerate your writing by
    • automating rote tasks (like writing the acknowledgements)
    • generating suggestions for better wording (like for article titles)
  • generate keywords and topic ideas (especially helpful for novices in a field)
  • improve writing by suggesting edits to your writing. It is very good at editing for
    • conciseness
    • organization
    • …but it sometimes changes the meaning so these edits should be taken as suggestions only

The Bad: GPT-4: AI does not have a way to access or search specific copyrighted or paywalled content on-demand. The AI model does not reproduce the entirety of copyrighted works or paywalled content but rather synthesizes the patterns and knowledge it has gained from various sources to generate human-like text.

The Ugly: Despite GPT-4’s capabilities, it maintains a tendency to:

  • make up facts and nonsensical content
  • double-down on incorrect information
  • perform tasks incorrectly
  • share your search history with strangers

What you can do to keep up…

  • PLAY with ChatGPT
  • TALK to your students and colleagues
  • READ/LISTEN/ ATTEND events (like this one sponsored by DLI)
  • EMBRACE & ADVOCATE for AI/algorithm literacy

 


This entry was posted on Wednesday, March 29th, 2023 at 3:34 pm and is filed under Resources. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply

Your email address will not be published. Required fields are marked *