Peter Johnson
@libraryvines.bsky.social
Medical librarian #medlibs also into Tolkien, history, Star Wars, amateur Excel skills, and inclusivity
“Treating [AI] with a hands-clean purist mindset will be extremely difficult and as activism, more alienating than effective. These are genuinely useful tools, and pretending they aren’t will not in fact win many hearts and minds.”
aredridel.dinhe.net/2026/02/12/t...
aredridel.dinhe.net/2026/02/12/t...
The AI hater's guide to code with LLMs (The Overview) | All Confirmation Bias, All The Time
aredridel.dinhe.net
February 13, 2026 at 8:48 PM
“Treating [AI] with a hands-clean purist mindset will be extremely difficult and as activism, more alienating than effective. These are genuinely useful tools, and pretending they aren’t will not in fact win many hearts and minds.”
aredridel.dinhe.net/2026/02/12/t...
aredridel.dinhe.net/2026/02/12/t...
Reposted by Peter Johnson
Do you know how badly you have to fuck up to create a remote execution vulnerability in notepad.exe
February 12, 2026 at 6:29 PM
Do you know how badly you have to fuck up to create a remote execution vulnerability in notepad.exe
I also stress in my AI information literacy instruction sessions: what these models generate is not “information” - to have that would require a deliberate human led transformation of the content.
LLMs do not exist outside of receiving a sequence of tokens in and returning with tokens out. They do not think, they do not exist in time outside of that input-output cycle, they have no consciousness to grow and reflect
Robin Sloan expresses it better than me www.robinsloan.com/lab/language...
Robin Sloan expresses it better than me www.robinsloan.com/lab/language...
February 11, 2026 at 3:34 PM
I also stress in my AI information literacy instruction sessions: what these models generate is not “information” - to have that would require a deliberate human led transformation of the content.
I also wonder how these models compare with both their more recent and more high powered versions. For the moment the timeliness context continues to be a challenge in assessing LLM capabilities.
"In our work, we found that none of the tested language models were ready for deployment in direct patient care."
#medlibs
www.nature.com/articles/s41...
#medlibs
www.nature.com/articles/s41...
Reliability of LLMs as medical assistants for the general public: a randomized preregistered study - Nature Medicine
In a randomized controlled study involving 1,298 participants from a general sample, performance of humans when assisted by a large language model (LLM) was sensibly inferior to that of the LLM alone ...
www.nature.com
February 11, 2026 at 2:56 PM
I also wonder how these models compare with both their more recent and more high powered versions. For the moment the timeliness context continues to be a challenge in assessing LLM capabilities.
Reposted by Peter Johnson
I need faculty to get on board with helping residents, post docs, phd students, masters students etc understand that you can only do a systematic review in less than 6 months if (a) there's only a tiny amount of evidence and (b) you're not doing anything else for those 6 months.
February 10, 2026 at 11:56 PM
I need faculty to get on board with helping residents, post docs, phd students, masters students etc understand that you can only do a systematic review in less than 6 months if (a) there's only a tiny amount of evidence and (b) you're not doing anything else for those 6 months.
Reposted by Peter Johnson
Reposted by Peter Johnson
We don’t talk enough about Alfre Woodard’s performance in Star Trek: First Contact. Stunning. And among other things, brought out some of Patrick Stewart’s moments. He rarely had a partner who could meet his energy, but she did. 🖖🏽
a woman is standing in front of a display of ships and says " you broke your little ships "
Alt: a woman is standing in front of a display of ships and says " you broke your little ships "
media.tenor.com
February 4, 2026 at 10:22 PM
We don’t talk enough about Alfre Woodard’s performance in Star Trek: First Contact. Stunning. And among other things, brought out some of Patrick Stewart’s moments. He rarely had a partner who could meet his energy, but she did. 🖖🏽
Reposted by Peter Johnson
Reproducibility and open science are great, but they don't necessarily equate to rigor. You can perfectly share a study and still draw weak conclusions. True rigor lives in the questions we ask, the designs we choose, and the inferences we make.
February 2, 2026 at 5:15 PM
Reproducibility and open science are great, but they don't necessarily equate to rigor. You can perfectly share a study and still draw weak conclusions. True rigor lives in the questions we ask, the designs we choose, and the inferences we make.
Reposted by Peter Johnson
googling “how to intimidate groundhog”
February 2, 2026 at 11:56 AM
googling “how to intimidate groundhog”
#medlibs and #evidencesynthesis nerds - the next Library Evidence Synthesis Services Symposium (LESSS) is upon us and open for free registration 😊 sites.google.com/view/lesss/r...
LESSS 2026 - Register
Register Now for LESSS 2026
June 15-18, 2026
11:30am - 3:30pm
sites.google.com
January 30, 2026 at 7:28 PM
#medlibs and #evidencesynthesis nerds - the next Library Evidence Synthesis Services Symposium (LESSS) is upon us and open for free registration 😊 sites.google.com/view/lesss/r...
I used Claude to figure out where the option to turn off the AI Meeting Notes prompt is in Notion’s settings (it’s in Notifications).
Pitting one AI against another is how we win this thing, folks 😉
Pitting one AI against another is how we win this thing, folks 😉
January 28, 2026 at 7:17 PM
I used Claude to figure out where the option to turn off the AI Meeting Notes prompt is in Notion’s settings (it’s in Notifications).
Pitting one AI against another is how we win this thing, folks 😉
Pitting one AI against another is how we win this thing, folks 😉
Policy via signage is never my favorite option, but this reminder in our staff area always makes me smile. 🧼
January 28, 2026 at 2:06 PM
Policy via signage is never my favorite option, but this reminder in our staff area always makes me smile. 🧼
Reposted by Peter Johnson
in a real role reversal today i got to give a bandaid to a medical student at the reference desk.
January 27, 2026 at 7:25 PM
in a real role reversal today i got to give a bandaid to a medical student at the reference desk.
#medlibs - a must read thread, especially for the indexing nerds 🤓
Just when you thought it was safe to dive into MeSH terms...
I'm popping up a bit late to give a peek at what I've found for new #2026MeSH descriptors. Behold, a #medlibs and adjacent #expertsearching thread!
I'm popping up a bit late to give a peek at what I've found for new #2026MeSH descriptors. Behold, a #medlibs and adjacent #expertsearching thread!
January 21, 2026 at 11:29 PM
#medlibs - a must read thread, especially for the indexing nerds 🤓
Reposted by Peter Johnson
Meet Yuriko, who brings knowledge from the past into the present so more people can learn from it #Wikipedia25
January 18, 2026 at 2:57 PM
Meet Yuriko, who brings knowledge from the past into the present so more people can learn from it #Wikipedia25
Reposted by Peter Johnson
#medlibs Google AI Overviews put people at risk of harm with misleading health advice
www.theguardian.com/technology/2...
www.theguardian.com/technology/2...
Google AI Overviews put people at risk of harm with misleading health advice
Exclusive: Inaccurate information presented in summaries, Guardian investigation finds
www.theguardian.com
January 14, 2026 at 1:26 PM
#medlibs Google AI Overviews put people at risk of harm with misleading health advice
www.theguardian.com/technology/2...
www.theguardian.com/technology/2...
😊
The name Peter "rocks". Literally. It comes from the Greek word meaning "rock" and has been borne by saints, emperors, musicians, cartoon characters, and athletes 🧵⬇️ (1/5)
January 12, 2026 at 1:20 PM
😊
Reposted by Peter Johnson
#medlibs #MedlibProblems Muscle memory forcing you to capitalize ANDs AND ORs in your day-to-day (non-search) communication.
January 9, 2026 at 4:01 PM
#medlibs #MedlibProblems Muscle memory forcing you to capitalize ANDs AND ORs in your day-to-day (non-search) communication.
Proud to have been a part of this 🙂 jamanetwork.com/journals/jam...
Melatonin Use in Young Children
This systematic review examines the long-term safety and effectiveness associated with melatonin use among young children.
jamanetwork.com
January 2, 2026 at 5:21 PM
Proud to have been a part of this 🙂 jamanetwork.com/journals/jam...
Reposted by Peter Johnson
Welcome to the Public Domain, THE MALTESE FALCON (1930) 🦅💎
🕵️ Sam Spade is on the Case of the Disappearing Copyright as The Maltese Falcon novel joins the public domain.
Learn more ➡️ blog.archive.org/public-domai...
#PublicDomainDay #CopyrightFree
🕵️ Sam Spade is on the Case of the Disappearing Copyright as The Maltese Falcon novel joins the public domain.
Learn more ➡️ blog.archive.org/public-domai...
#PublicDomainDay #CopyrightFree
January 1, 2026 at 5:25 AM
Welcome to the Public Domain, THE MALTESE FALCON (1930) 🦅💎
🕵️ Sam Spade is on the Case of the Disappearing Copyright as The Maltese Falcon novel joins the public domain.
Learn more ➡️ blog.archive.org/public-domai...
#PublicDomainDay #CopyrightFree
🕵️ Sam Spade is on the Case of the Disappearing Copyright as The Maltese Falcon novel joins the public domain.
Learn more ➡️ blog.archive.org/public-domai...
#PublicDomainDay #CopyrightFree
Reposted by Peter Johnson
Essentially, I think that LLM-generated dialogue only makes sense if you believe that the product of writing is words.
But the product of writing isn't words, any more than the product of justice is prisoners, or the product of love is weddings. The product of writing is meaning.
But the product of writing isn't words, any more than the product of justice is prisoners, or the product of love is weddings. The product of writing is meaning.
I had five fascinating conversations for this RPS piece, gathering perspectives on genAI chatbots in games, as seen in Where Winds Meet. Thanks to @meghna.bsky.social, @pyrofoux.bsky.social, @maxkreminski.bsky.social, and @chrisgardiner.bsky.social www.rockpapershotgun.com/horrible-bor...
"Horrible", "boring", and "cheap": Experts pan new chatbot NPCs, but some leave room for optimism
We speak to a collection of developers and AI experts about the potential impact of LLM-powered NPCs in videogames.
www.rockpapershotgun.com
December 23, 2025 at 7:06 PM
Essentially, I think that LLM-generated dialogue only makes sense if you believe that the product of writing is words.
But the product of writing isn't words, any more than the product of justice is prisoners, or the product of love is weddings. The product of writing is meaning.
But the product of writing isn't words, any more than the product of justice is prisoners, or the product of love is weddings. The product of writing is meaning.
A listserv changed their config and has now bypassed the filter directing it to a subfolder and away from the inbox.
a black and white photo of a waterfall coming down a wall .
Alt: a black and white photo of a waterfall coming down a wall .
media.tenor.com
December 23, 2025 at 10:56 PM
A listserv changed their config and has now bypassed the filter directing it to a subfolder and away from the inbox.
Reposted by Peter Johnson
The [free!] recording is available for our December 15th Taubman Talks webinar "Developing Systematic Search Skills: Peer-to-Peer Learning with Practice Sets"!
You can find the recording and files at the bottom of the webinar page on the Taubman Talks site:
guides.lib.umich.edu/TaubmanTalks...
You can find the recording and files at the bottom of the webinar page on the Taubman Talks site:
guides.lib.umich.edu/TaubmanTalks...
Research Guides: Taubman Library Talks: Developing Systematic Search Skills: Peer-to-Peer Learning with Practice Sets (December 2025)
The home for videos and webinars created by Informationists at the Taubman Health Sciences Library
guides.lib.umich.edu
December 23, 2025 at 2:51 PM
The [free!] recording is available for our December 15th Taubman Talks webinar "Developing Systematic Search Skills: Peer-to-Peer Learning with Practice Sets"!
You can find the recording and files at the bottom of the webinar page on the Taubman Talks site:
guides.lib.umich.edu/TaubmanTalks...
You can find the recording and files at the bottom of the webinar page on the Taubman Talks site:
guides.lib.umich.edu/TaubmanTalks...
Reposted by Peter Johnson
New measure of academic impact just dropped: number of citations received to papers that chatgpt created and fraudulently attributed to you.
These are all the 42 papers Google Scholar has found citing "our" paper that we never wrote - who knows how much actual human touch went into them? scholar.google.com.vn/scholar?star...
scholar.google.com.vn
December 19, 2025 at 7:19 PM
New measure of academic impact just dropped: number of citations received to papers that chatgpt created and fraudulently attributed to you.