Miguel and I are currently working on a Databit article that explores how generative AI is being used in Information Management (IM) tasks. Our goal is to compile a collection of real-world cases and ideas that highlight the potential and current use of AI in the IM world (some of cases will extend into science).
To make this article as insightful and collaborative as possible, we’d like to ask for your input on two key questions:
Do you have any AI-related accomplishments in your IM tasks that you’d like to share?
(For example, tools you’ve used, creative ways you’ve applied AI, or any success stories you’d like to highlight.) @whiteaker I got your example :).
Are there parts of your IM-related tasks where you wish AI tools could help?
(This will help us identify areas to explore and provide direction for future AI tool research.)
We’d love to hear your thoughts, experiences, and ideas. Feel free to reply here or reach out directly. Your input will be invaluable for shaping this article and expanding our collective understanding of how AI can enhance IM work.
@lkui@Miguel_Leon the list of potential things we could speed up with AI is long!.
Preparing EML (?)
Spell/grammar/content checking EML (it is easy to validate for syntactically valid EML, but could I be use to provide feedback on what conceptual information is missing or to suggest edits such as removing manuscript results from dataset abstracts. Probably a pipe-dream…
Data qc
Data gap-filling
Data harmonization
Semantic annotation - even if this had to be checked by a human if AI could give a first pass of possible ontologies and specific terms based on content?
Improving dataset search
AI for species identification (using AI-labeled photo-quadrats to supplement field work by taxonomic specialists)
Hello Ikui and Migual,
I would offer two thoughts.
First that AI has been around since the very first computers as they can process ‘data’ much faster than any human. Thus computers have taken control in more ways than one.
However as E.O. Wilson says in his seminal book Consilience - "We are drowning in data while starving for wisdom’. How do you or any scientists seperate the data and identify the wisdom first? I have yet to get a cogent response to this basic question. It goes to the heart of what is socially relevant research and science.
I’m happy to engage in this question as I’ve tried to bring it up for decaes now.
I’ve made modest use of GitHub Co-pilot in R programming tasks. As I code it makes suggestions about what it thinks I’ll want to do next, that I can accept, edit or ignore as appropriate.
I’ve found it especially helpful if there are multiple similar tasks (e.g, converting multiple variables from feet to meters). Sometimes it is way off base, but I am also sometimes astonished about how it guessed what I wanted to do next.