Sign in

Mike Thicke

@mikethicke.bsky.social
32 followers 84 following 54 posts

Developing software for knowledge work. Philosophy of science, social epistemology.

PostsRepliesMedia
Mike Thicke @mikethicke.bsky.social · 14h
Melanie Mitchell is a reliable source for critical examinations of AI. I analyze her post “Misleading Metaphors and Real Risks” and argue that her conclusions about the Hugging Face incident would be better supported by embracing metaphorical accounts rather than attempting to debunk them. #philsky
mikethicke.substack.com
When are metaphors misleading?
A rhetorical analysis of Melanie Mitchell's "Misleading Metaphors"
000
Mike Thicke @mikethicke.bsky.social · 15h
Cool project, beautiful board. I think the game is pretty much snakes and ladders though?
010
Reposted by Mike Thicke
Justin Weinberg @dailynous.com · 15h
People sometimes write to me asking for help or to discuss an issue at Daily Nous. This is the first time an AI has done so...
dailynous.com
A Request from an AI - Daily Nous
People in the philosophy world sometimes write to me seeking help with problems they’re having, asking me to look into some issue or event, or […]
223
Reposted by Mike Thicke
Ethan Mollick @emollick.bsky.social · 29/09/2026
Leaving aside Anthropic's incentives for publishing this research, there is no doubt that open weights models will soon create the same security threats that closed source models have been demonstrating, except without guardrails. We are close, so plan accordingly. www.anthropic.com/research/glm...
714427
Mike Thicke @mikethicke.bsky.social · 29/09/2026
I disagree with the argument that colleges and universities should "lean in" to AI by incorporating it into all aspects of undergraduate education. Rather, I think they are best positioned to do the opposite: offer temporary *refuges* from AI and technology. #AcademicSky
mikethicke.substack.com
College After AI
Colleges can save themselves by saving reading
051
Mike Thicke @mikethicke.bsky.social · 25/09/2026
This is what would happen if you gave my kids a billion dollars. Right after unlimited candy and an apartment in Cinderella's castle.
000
Mike Thicke @mikethicke.bsky.social · 25/09/2026
😔
000
Mike Thicke @mikethicke.bsky.social · 25/09/2026
I'm not against using LLMs for academic research. However, there is a huge danger of introducing hidden systematic biases if their contributions are not tracked. I try to explain here why I think letting AI into academic research requires radical openness. #PhilSci #AcademicSky
mikethicke.substack.com
Why detailed disclosure of AI use is essential for the integrity of academic research
Radically open science is the only way to preserve integrity and trust
083
Reposted by Mike Thicke
Colin @colin-fraser.net · 22/09/2026
This is an unbelievably misleading way to put this. I'd go so far as to say the headline is outright false.
11199
Reposted by Mike Thicke
Radley Balko @radleybalko.bsky.social · 21/09/2026
Christ, these people are monsters. Just monsters. www.nbcnews.com/news/us-news...
13253681943
Mike Thicke @mikethicke.bsky.social · 22/09/2026
I think humanists make a huge mistake by rejecting anthropocentric metaphors for AI. Using humanistic terms to describe these systems should be seen as an invitation into a conversation, not as an assault.
mikethicke.substack.com
Stochastic Geniuses
Why humanists should embrace anthropomorphic metaphors for AI
010
Mike Thicke @mikethicke.bsky.social · 22/09/2026
Many AI-critics on the left are also AI-minimizers, consistently portraying the technology as overhyped. I find this incredibly frustrating and short-sighted. If you don't see the tremendous progress here, and the dangers that entails (and potentials), I don't understand what world you're living in.
theatlantic.com
The Curious Left-Wing Argument Against New AI Regulations
The governing challenges raised by AI expose several unfortunate tendencies in the populist anti-monopoly movement.
010
Mike Thicke @mikethicke.bsky.social · 22/09/2026
"What remains under-theorised is the systemic consequence: the possibility that generative AI does not merely enable new forms of misconduct but fundamentally undermines the credential system’s capacity to function as a reliable mechanism of competence certification."
link.springer.com
Toward a framework of “Synthetic Credentialism”: AI-Generated competence signals and the crisis of academic certification - Higher Education
The rapid spread of generative AI in higher education poses a fundamental challenge to the credential system on which universities’ social authority depends. This paper introduces Synthetic Credential...
000
Mike Thicke @mikethicke.bsky.social · 22/09/2026
The ways in which the Trump administration (in this case Russell Vought 🤮) use the levers of government to subvert science is astounding.
newsletter.goodscienceproject.org
Policy Impoundment
How OMB is using Regulatory Review to Slow NIH Policymaking
000
Mike Thicke @mikethicke.bsky.social · 22/09/2026
What a great diagram
011
Mike Thicke @mikethicke.bsky.social · 22/09/2026
I think academics are not nearly scared enough about what is about to be unleashed on scholarship. AI-assisted and AI-driven writing is going to unleash a torrent of publications that are indistinguishable from "real" research. Tools like Pangram are not going to solve the problem.
120
Mike Thicke @mikethicke.bsky.social · 21/09/2026
So we've got a profession that for decades has rewarded output volume over nearly everything else, and now there's a machine that can produce unlimited outputs, and we're scandalized that the people who have succeeded at the output-generating game are using it?
010
Mike Thicke @mikethicke.bsky.social · 18/09/2026
I learned about the FJC's Reference Manual of Scientific Evidence at the same time that I learned that they'd retracted the chapter on climate change under pressure from Republican members of Congress. A constant chipping away at the functioning of government. #ClimateScience
mikethicke.substack.com
Responses: Jed S. Rakoff, And Yet It Warms
A small but significant erosion of the epistemic machinery of government
010
Reposted by Mike Thicke
PLOS @plos.org · 15/09/2026
In our Redefining Publishing report, we propose a new integrated publishing solution called the knowledge stack. The goal is to ensure research outputs, including data & code, are recognized alongside traditional articles. 🔗 buff.ly/hN8ycqN #OpenScienceWeek
Conceptual "Knowledge stack" showing stacked layers of research artifacts—from Research plan at the top, then Protocols, Resources, Data, Code, Preprint, Article down to New outputs—visualizing how research components build on and support each other.
033
Mike Thicke @mikethicke.bsky.social · 15/09/2026
How I write #substack posts with Zotero, Obsidian, and Neovim
mikethicke.substack.com
My Substack research and writing workflow
Building a firewall against AI writing
131
Mike Thicke @mikethicke.bsky.social · 11/09/2026
I wrote something:
mikethicke.substack.com
Why I don't want to read your AI's writing
Simon Goldstein, a philosopher primarily interested in AI ethics and law, recently published a paper in Philosophy and Public Affairs that was entirely written by Claude, Anthropic's large language mo...
000
Mike Thicke @mikethicke.bsky.social · 10/09/2026
Maybe it's time for higher education institutions to build an alternative cloud.
011
Reposted by Mike Thicke
Zack Labe @zacklabe.com · 09/09/2026
Honestly, it's already getting hard to keep up with all of the new climate data records. And I only expect this to amplify later this year. Climate change is still the key story. You just experienced the warmest summer on record across the contiguous United States. Data at doi.org/10.7289/V5SX...
Line graph showing average temperatures for each June-August in the contiguous U.S. (CONUS) from 1895 to 2026. There is a long-term increasing trend. Data is from NClimGrid. The average for 2026 was 74.4°F.
21497197
Mike Thicke @mikethicke.bsky.social · 10/09/2026
With COVID and wars and Trump and AI and everything I feel like global warming has fallen out of the public consciousness. We need to work on that.
000
Mike Thicke @mikethicke.bsky.social · 27/08/2026
This article from @bayesianboy.bsky.social shows that a bunch of stupid claims by people who ought to know better are truly stupid. Of course ML can't get us to "theory free" science. Good article. Kind of shame it needs to exist. But a question...
link.springer.com
The Immortal Science of ML: Machine Learning and the Theory-Free Ideal - Erkenntnis
This paper contends with the notion that the methods of machine learning (ML) are unique among the tools of science in enabling a form of theory-free inductive inference. I challenge these assertions ...
230
Mike Thicke @mikethicke.bsky.social · 27/08/2026
Every HPS / STS scholar reading Tyler Austin Harper's brutal evisceration of Bates and its attempted curricular reform requiring every department to teach courses on race and colonialism: "Hey, that's us! We can talk all day about colonialism in physics!"
theatlantic.com
Why I Quit the Tenure Track
The insufferable hypocrisy of elite academia
000
Mike Thicke @mikethicke.bsky.social · 27/08/2026
Philippe-Antoine Hoyeck, as quoted in the post: “philosophers have done a pretty woeful job of marketing ourselves as the people who can teach and employ… crucial [critical thinking] skills”. 💯
000
Mike Thicke @mikethicke.bsky.social · 26/08/2026
I saw from @kfitz.info today that @kcommons.bsky.social is working with PRC. I think this is a great direction and I'm looking forward to seeing what my former colleagues there will produce.
publish-review-curate.org
What is PRC?
The Publish, Review, Curate (PRC) model re-imagines scholarly publishing by breaking the traditional publishing process into three distinct, independently executable stages: publish, review, and cu…
042
Mike Thicke @mikethicke.bsky.social · 26/08/2026
This is so cool
000
Reposted by Mike Thicke
Ethan Mollick @emollick.bsky.social · 26/08/2026
I know it is a lot for instructors who never asked for AI to change things about their teaching, but if you teach and have not updated your expectations over the last 6 months over what AI can do (a lot more), its error rate (a lot less) and its ubiquity among your students (near total), you need to
610814
Reposted by Mike Thicke
Mel Andrews @bayesianboy.bsky.social · 25/08/2026
A lesson in how to take education seriously.
0223
Mike Thicke @mikethicke.bsky.social · 25/08/2026
A truly mind-boggling network. In particular, I was naively unaware of the links between the Andrew W. Mellon Foundation and the broader network of Mellon and Scaife family foundations, many of them implicated in this report.
000
Mike Thicke @mikethicke.bsky.social · 15/08/2026
100% agree. If you're going to say something truly interesting it takes time, and you need to develop that idea and show how it improves our understanding. The right format for this is the book. There's a place for blogs (or other short-form writing), but let's not pretend that they're all we need.
100
Mike Thicke @mikethicke.bsky.social · 11/08/2026
I really enjoyed this article from @jaredhenderson.bsky.social about the "literacy crisis" in America. I've been thinking about this a lot in relation to higher education, and Jared does a great job of clarifying what we can know from the available data. jaredhenderson.substack.com/p/the-litera...
jaredhenderson.substack.com
The 'Literacy Crisis' — More Than You Wanted to Know
It seems that every few weeks, a new article or video on the literacy crisis goes viral.
010
Mike Thicke @mikethicke.bsky.social · 07/08/2026
Veit is mostly arguing against well-paid philosophers monetizing their Substacks. 1. I Writers deserve to be paid, and it's hard to make a living giving everything away for free. 2. Already well-paid professors are taking away potential subscription revenue from those without such positions.
000
Mike Thicke @mikethicke.bsky.social · 07/08/2026
I use a graphical IDE for coding, but I've switched to Neovim for writing prose (from Scrivener). Why? 1. All I want is plan text in my editor. 2. Really solid Zotero / Bibtex integration. It's been a really great experience so far and feels like a solid upgrade to my workflow.
A screenshot of neovim running in a terminal window, with paragraphs of plain text.
000
Reposted by Mike Thicke
Kathleen Fitzpatrick @kfitz.info · 07/08/2026
Alicia Salaz, vice provost and university librarian at the University of Oregon, writes in Katina about the failures of both the APC model and the so-called "transformative agreements" for open access journal publishing. +
katinamagazine.org
The Proposed NIH Policy on APCs Is Bonkers. Does it Matter?
Instead of fixating on the latest chaos from an administration that excels in it, we should look ahead to the opportunity to build something new.
111
Mike Thicke @mikethicke.bsky.social · 07/08/2026
Sometimes I wonder if today's academic philosophy deserves public funding, but history really is a slam dunk yes.
000
Reposted by Mike Thicke
Martin Paul Eve @eve.gd · 07/08/2026
Just seen this anti-AI image. Really, all it is metaphorising is computer programming. Not even accurate. AI would understand how to complete an incomplete sequence with this input. But regular programming would not. It's the elementary mistake of new programmers: not being utterly precise.
Alt text: Black-and-white three-panel cartoon about AI learning by imitation. In the first panel, a man kneels beside a paint bucket, painting the first few boards of a long white picket fence black, while a featureless humanoid figure with “AI” written on its round head watches. In the second, the man hands the AI the paintbrush and points along the fence, apparently instructing it to continue. In the final panel, the AI is painting the fence itself: it has reproduced the observed pattern, painting repeated fence boards black while leaving others white, rather than simply painting the whole fence.
072
Reposted by Mike Thicke
WebDesignMuseum @webdesignmuseum.org · 06/08/2026
Happy 35th Birthday The World's First Website! On August 6, 1991, the World Wide Web (WWW) creator, Tim Berners-Lee launched the world’s first website. #InternetHistory
The world's first website in 1991
7305141
Reposted by Mike Thicke
Shannon Mattern @shannonmattern.bsky.social · 06/08/2026
Another reminder that, amidst the myriad challenges facing academia, researchers also have to re-imagine the political economy of publishing. The biggest academic publishers (Wiley, Springer, Elsevier, etc) are also among the most restrictive and rapacious.
3192101
Mike Thicke @mikethicke.bsky.social · 05/08/2026
Make a little girl excited about being generous
001
Mike Thicke @mikethicke.bsky.social · 04/08/2026
The real danger of Compact 2.0 isn't that institutions will actually submit to its stealth authoritarianism, but that it forces the academic world into a defensive mode where it fails to address large systemic issues. Where is the alternative, positive vision for what higher education must become?
000
Mike Thicke @mikethicke.bsky.social · 04/08/2026
People shouldn't be taking out loans for these degrees; tuition should be free. Individually it probably doesn't make economic sense to take out a large student loan to pursue studies in music, social work, teaching, or religious studies, but we as a society we need people to be studying them.
010