johnsontoddr.bsky.social
Professor of Biomedical Informatics.
25 posts
15 followers
53 following
Getting Started
Active Commenter
comment in response to
post
Things are not getting better. I'm really concerned about applications of this technology. We have already seen terrible consequences of opioid, parole, health claims, etc. when using ML. Students are being accused of using AI to write, when they are not.
comment in response to
post
Sounds about right for the Trump admin. We are in for a terrible few years. At least, I hope it is just a few. I'm not sure whether it will be better or worse being in Texas, but I suspect worse.
comment in response to
post
Its really nice that a vendor set up to sell them food as they arrived ;-)
comment in response to
post
Also, one of my students had a submission falsely flagged as being done with AI and had to have a formal meeting to defend their work. This also happened to one of my daughters in high school about a year ago.
comment in response to
post
True story: I had a nightmare last night where all research, grading, and faculty student evals were being done by Gen AI systems. I kept pointing out to admin and other faculty that the systems were just making things up, but I was ignored because the AI was more efficient!
comment in response to
post
I have flip down step for the bed of my truck that got stuck. Google AI told me to get my truck going at least 5 mph then run to the back of the truck to push the step down!
comment in response to
post
I now ignore Google's AI summary. It is not nearly reliable enough and/or informative enough to even waste time reading it.
comment in response to
post
I'm on the 7th journal submission for a paper. I keep putting off the changes because the latest reviews are way off base and the paper was far better 3 journal submissions ago. The review process is way too idiosyncratic.
comment in response to
post
I think this is what we call MAGA Logic. I keep hearing people voted for Trump because of inflation. Will they finally see their mistake once the prices start to skyrocket? Somehow I doubt it.
comment in response to
post
Turns out Hilary was actually being kind when she used the term "deplorables."
comment in response to
post
I have trouble convincing customers to spend $200 on a grinder. I make coffee for our research meetings using my own equipment in my office: a moccamaster and a Fellow Opus. I use coffee I roast myself, which means it is sometimes experimental. The coffee in the faculty/student kitchen is horrid.
comment in response to
post
Clearly they skipped Espresso 101. I didn't. You can see me here: roothedaycoffee.com
comment in response to
post
Regulations and NIMBYism against high density and low income housing are a big part of the problem as well. We need regulatory and institutional reform in this country at nearly all levels. Ezra Klein has a book coming out about these issues.
comment in response to
post
Yes, my family is finally all on board with moving out of Texas, of course job for me is a big issue in ability to move. I suspect that we would see more moves out of repressive states if it was easier to find jobs.
comment in response to
post
Probably a better chance of winning a bet that another AI Winter is coming.
comment in response to
post
How do we do that when that group will take anything as a talking point. To be fair, there are already some guidelines on doing and reporting obs epi/health studies and there are more in the works, such as for target trial emulation.
comment in response to
post
Keep up the battle. With Dunn and Wilks buying out GOP politicians, we need all the help we can get.
comment in response to
post
Medical research has an issue here, because most papers using patient data make neither the data, nor the code used to analyze it available. The publish or perish nature of academia (and the big pharma incentive) create more problems. Replication is a big problem.
comment in response to
post
Nice to see this. At least in my area, it will break the monotony of talks and papers using LLMs to make incremental (if any) improvements to biomedical informatics tasks.
comment in response to
post
Both Soar and ACT-R are inherently symbolic, though ACT-R has an underlying statistical component. But they represent the more symbolic, deliberative models of cognition, not the "subsymbolic". Embeddings and DL perceptual models could augment their basic framework.
comment in response to
post
I see two areas: (1) using what we have in a collaborative fashion with humans, similar to a several year old distill paper out of Google; and (2) exploring use of LLMs and related tech as parts of cognitive architectures ala SOAR and ACT-R.
comment in response to
post
Perhaps another AI Winter? Though I suspect we will have a long Spring, since most of our PhD students now seem to be giving presentations with titles like "Application of LLMs to... [pick your favorite ML task in healthcare]"
comment in response to
post
Count me in. Just came from fasXist.