AI and the CMS

[Jegelewicz]

From TDWG Slack

I’m interested in hearing what people think the future role of llms like gpt-4 might be in curating and developing the darwin core standard and extensions. I’ve tried asking it questions that people ask me (e.g. how should i describe plant cover in percentage and subplot frequency for vegetation data), and asked to how it would structure a new dwc extension to incorporate some raw data, and it made some very good suggestions. I think there’s a lot of exciting potential here.

It is definitely pretty-darn-good at mapping free text to Darwin Core. At TaxonWorks we're looking toward having this in our Digitzation Workflow so that you could give it text, it would suggest what goes where and a data technician decides if correct to go into the database. People at NHM (I'm thinking about Steen Dupont) are working on the image AI. I think it's only going to improve and we're going to think of more possibilities.