karin verspoor (professor)

Executive Dean @RMITComputing; AI in medicine #AAAiH; text mining, bioinformatics, #DigitalHealth #NLProc #techdiversity #TeamHB3 pronouns: she/her

[karin verspoor (professor) @karinv]

I have worked on #NLProc for ~30 years; until recently most people did not understand or appreciate the nature of my work. Then came #ChatGPT.

I have been sharing some thoughts on this technology, and I am pleased to share them with you too!

https://t.co/RqRC1O3w3R

[karin verspoor (professor) @karinv]

@DrShaneRRR “Science does not have a communication problem. The government has a foresight problem.”
Well said!

[Sholto David @addictedtoigno1]

A new video for my YouTube channel, this one is about image manipulation in a Nature Neuroscience Paper by researchers at the University of Michigan.

https://t.co/h7y7ZlU7Rk

[karin verspoor (professor) @karinv]

@fake_journals My approach: look at who cited papers. Start with a recently published paper with a high citation count. Take this 2025 paper with 13 citations. Look at "Cited by". Almost entirely author self-citations or highly nonspecific citations. A self-reinforcing citation network?

[karin verspoor (professor) @karinv]

@Duke07696804 @fake_journals This is already happening!

[karin verspoor (professor) @karinv]

@TobyWalsh You and me both!

[Isaac Kohane @zakkohane]

The incidentalome continues to grow and cost $$$$ https://t.co/bjLQ35npdv

[karin verspoor (professor) @karinv]

@JulioGonzalo1 @sansalido @guillermomarco_ Here also we created a targeted test suite for negation. Controlled variation of examples gives insight into capabilities that a large benchmark cannot.
https://t.co/mtWxNvAEpj

[karin verspoor (professor) @karinv]

@JulioGonzalo1 We had very consistent results when introducing negation!

https://t.co/TIlkvnU8uU

[karin verspoor (professor) @karinv]

@RMITComputing undergraduate student Milindi Kodikara presenting a paper based on her Honours project with me. She also graduated with First Class Honours for her Honours year.

Well done, Milindi!

[Australasian Language Technology Association @altanlp]

Milindi Kodikara from @RMIT is presenting the long paper today at #ALTA2024 titled "Lesser the Shots, Higher the #Hallucinations: Exploration of Genetic Information Extraction using Generative Large Language Models", evaluating #LLMs on extraction of genetic information.
#NLP

[karin verspoor (professor) @karinv]

@lpachter Do you have a sense of whether this was a true “innovation” or just a better information retrieval strategy? ie was it known just not to you?

[karin verspoor (professor) @karinv]

@AvpElk It’s precisely because of how they are trained, and follows from reinforcing the statistical patterns in the data.

[karin verspoor (professor) @karinv]

@AvpElk @ylecun No. You will end up with model collapse.

https://t.co/wf2ys525h4

[karin verspoor (professor) @karinv]

@aadi_joshi @AU_NSWPF That’s horrible. I’m sure it was frightening. I’m relieved that he stopped at verbal abuse.

[karin verspoor (professor) @karinv]

@elizlaraki WTF indeed. Unintended consequences of AI automation …

[karin verspoor (professor) @karinv]

@innovator_lc @ipfconline1 @ConversationEDU @jblefevre60 @Khulood_Almani @ahier I completely agree! That is the sweet spot, and where my own research has focused.

We made the same argument for clinical AI/NLP a few years ago. Support information needs, don’t replace critical thinking.

https://t.co/cZMuTSjk8S

[karin verspoor (professor) @karinv]

@DrShaneRRR “scarred” not “scared”! Well maybe scared too but seriously they were just … disturbing. At least as I recall.

[karin verspoor (professor) @karinv]

As I said at #WCRI2024 @WCRIFoundation and last week at @ANU_CPAS, #GenAI will enable bad actors in science to accelerate their contamination of scientific literature.

This has broad repercussions for trust in science, and there is a significant human toll, as reflected here.

[Calli McMurray @callimcflurry]

Science is built on trust. What happens when someone destroys it?

My first feature for @_TheTransmitter investigates the emotional and existential fallout for one lab in the aftermath of a misconduct case.
https://t.co/xawx3J7mCC

[karin verspoor (professor) @karinv]

@OlegZendel I’m confused why a LLM is needed for this use case?

[karin verspoor (professor) @karinv]

@ARC_Tracker This is particularly troubling as AFAIK the sector has grown substantially in that time! This does not bode well for academic career progression.