Alex P. Miller
@alexmiller.phd
Reposted by Alex P. Miller
Academia will form these little pockets -- people whose theorizing is outrageous & supported by methods outdated since the 90s -- but once it reaches a critical size those people just review each others papers & grants, form societies, hand out awards etc, like a self-contained parallel society.
June 3, 2025 at 5:31 AM
Academia will form these little pockets -- people whose theorizing is outrageous & supported by methods outdated since the 90s -- but once it reaches a critical size those people just review each others papers & grants, form societies, hand out awards etc, like a self-contained parallel society.
Reposted by Alex P. Miller
This article is now predictably popular on Bsky but key parts of it are, well, hallucinations.
They try to claim "errors are rising" on "new reasoning systems" primarily based on Vectara hallucination leaderboard and one OpenAI document. So let's look what those actually show.
They try to claim "errors are rising" on "new reasoning systems" primarily based on Vectara hallucination leaderboard and one OpenAI document. So let's look what those actually show.
The newest and most powerful A.I. technologies — so-called reasoning systems from companies like OpenAI, Google and the Chinese start-up DeepSeek — are generating more errors, not fewer. As their math skills have notably improved, their handle on facts has gotten shakier.
A.I. Hallucinations Are Getting Worse, Even as New Systems Become More Powerful
A new wave of “reasoning” systems from companies like OpenAI is producing incorrect information more often. Even the companies don’t know why.
www.nytimes.com
May 6, 2025 at 3:57 PM
This article is now predictably popular on Bsky but key parts of it are, well, hallucinations.
They try to claim "errors are rising" on "new reasoning systems" primarily based on Vectara hallucination leaderboard and one OpenAI document. So let's look what those actually show.
They try to claim "errors are rising" on "new reasoning systems" primarily based on Vectara hallucination leaderboard and one OpenAI document. So let's look what those actually show.
One of those things that says what it doesn't say by saying it
The fact this even has to be said out loud is already a bridge too far. Our boy is really working on his seventh bankruptcy.
May 7, 2025 at 12:26 AM
One of those things that says what it doesn't say by saying it
Vibe coding is going to re-open the web. It's never been easier to make an interesting website, page, or single page app.
April 28, 2025 at 9:39 PM
Vibe coding is going to re-open the web. It's never been easier to make an interesting website, page, or single page app.
Posts like this on reddit all the time:
April 28, 2025 at 6:16 PM
Posts like this on reddit all the time:
A long but and worthwhile essay on where we are with technocracy/expertise and populism and how difficult the problems we face really are. Too few academics seriously think about their role as experts in society, apart from the ways that position serves their personal career.
Weaponization of Expertise
Some Thoughts on Elites, Populism, and The Mess We Are All In
rbnmckenna86.substack.com
April 25, 2025 at 11:20 PM
A long but and worthwhile essay on where we are with technocracy/expertise and populism and how difficult the problems we face really are. Too few academics seriously think about their role as experts in society, apart from the ways that position serves their personal career.
On Bayesian background priors, the long trail of failures in health tech (RIP Theranos) give me pause about companies making similar claims. Many people also get their blood tested in the course of routine doctor visits. It's not that hard to upload bloodwork to <leading general purpose model>.
Superpower wants to help people detect and address health issues before symptoms appear | TechCrunch
A startup called Superpower is publicly launching what it calls the world’s first super app that is designed to help people better understand and take
techcrunch.com
April 25, 2025 at 4:52 PM
On Bayesian background priors, the long trail of failures in health tech (RIP Theranos) give me pause about companies making similar claims. Many people also get their blood tested in the course of routine doctor visits. It's not that hard to upload bloodwork to <leading general purpose model>.
Troubleshooting AI systems is hard, because sometimes the AI solves its own problem, and sometimes it doesn't. This makes the fundamental task of measuring things harder.
April 23, 2025 at 4:56 AM
Troubleshooting AI systems is hard, because sometimes the AI solves its own problem, and sometimes it doesn't. This makes the fundamental task of measuring things harder.
Let's try Google's new "AI Mode"... Right on the splash page, they suggest an interesting query: "houseplants that improve air quality and don't need much light". (1st image)
I went ahead and clicked it and was given a long reply with extensive citations (2nd image).
Very cool! .... right?
I went ahead and clicked it and was given a long reply with extensive citations (2nd image).
Very cool! .... right?
March 31, 2025 at 6:36 PM
Let's try Google's new "AI Mode"... Right on the splash page, they suggest an interesting query: "houseplants that improve air quality and don't need much light". (1st image)
I went ahead and clicked it and was given a long reply with extensive citations (2nd image).
Very cool! .... right?
I went ahead and clicked it and was given a long reply with extensive citations (2nd image).
Very cool! .... right?
First time in basically ever I've been added to a long (10,000+) queue for the archive.today. Pages are super slow to load too. Noting alongside reports of serious uptick in web-enabled bots and crawlers recently.
Open Source devs say AI crawlers dominate traffic, forcing blocks on entire countries
AI bots hungry for data are taking down FOSS sites by accident, but humans are fighting back.
arstechnica.com
March 26, 2025 at 6:26 AM
First time in basically ever I've been added to a long (10,000+) queue for the archive.today. Pages are super slow to load too. Noting alongside reports of serious uptick in web-enabled bots and crawlers recently.
American policy is pretty easy to understand right now. Trump desperately aspires to be a “great man”. One of his simplistic views of history is that great men conquer foreign lands and grow their empire.
One man’s ego, and hundreds of cowards in congress who fail to stop him, explains it all.
One man’s ego, and hundreds of cowards in congress who fail to stop him, explains it all.
Usha Vance, the second lady, is scheduled to join the White House national security adviser, the energy secretary and other U.S. officials to visit Greenland this week, amid President Trump’s continued push to take over the island, officials said on Sunday.
U.S. Officials to Visit Greenland Amid Trump’s Talk of a Takeover
Usha Vance, the second lady, will attend cultural events while Michael Waltz, the national security adviser, visits troops stationed on the island that President Trump has said the United States will ...
www.nytimes.com
March 23, 2025 at 6:44 PM
American policy is pretty easy to understand right now. Trump desperately aspires to be a “great man”. One of his simplistic views of history is that great men conquer foreign lands and grow their empire.
One man’s ego, and hundreds of cowards in congress who fail to stop him, explains it all.
One man’s ego, and hundreds of cowards in congress who fail to stop him, explains it all.
Reposted by Alex P. Miller
I don't think people understand the *massive* gulf between fire injuries/deaths and traffic violence.
Here's an example:
Average annual fire injures in Berkeley: 2
Average annual traffic injuries in Berkeley: 694
Here's an example:
Average annual fire injures in Berkeley: 2
Average annual traffic injuries in Berkeley: 694
March 17, 2025 at 8:58 PM
I don't think people understand the *massive* gulf between fire injuries/deaths and traffic violence.
Here's an example:
Average annual fire injures in Berkeley: 2
Average annual traffic injuries in Berkeley: 694
Here's an example:
Average annual fire injures in Berkeley: 2
Average annual traffic injuries in Berkeley: 694
The first principal component of the internet (or perhaps all mass media) is attention. Intention is a rounding error.
"Now don't harass anyone mentioned in this video!!"
the fact that so many Influencers(tm) feel the need to say that says a lot about both their audience and the fact they publish anyway
the fact that so many Influencers(tm) feel the need to say that says a lot about both their audience and the fact they publish anyway
March 15, 2025 at 9:22 AM
The first principal component of the internet (or perhaps all mass media) is attention. Intention is a rounding error.
We don’t need to outlaw cars in America. We merely need to legalize human-centric development in the highest density regions.
“The right to have access to every building in the city by private motorcar, in an age when everyone possesses such a vehicle, is actually the right to destroy the city.”
- Lewis Mumford
- Lewis Mumford
March 11, 2025 at 4:26 AM
We don’t need to outlaw cars in America. We merely need to legalize human-centric development in the highest density regions.
So many harmful tropes of the dominant academic culture on display here. If you have editorial/institutional power and you’re not trying to address this problem, you are doing a great disservice to science and young scientists.
“A Post Mortem on the Gino Case”: “Committing fraud is, right now, a viable career strategy that can propel you at the top of the academic world.”
statmodeling.stat.columbia.edu/2025/03/08/a...
statmodeling.stat.columbia.edu/2025/03/08/a...
“A Post Mortem on the Gino Case”: “Committing fraud is, right now, a viable career strategy that can propel you at the top of the academic world.” | Statistical Modeling, Causal Inference, and Socia...
statmodeling.stat.columbia.edu
March 8, 2025 at 3:04 PM
So many harmful tropes of the dominant academic culture on display here. If you have editorial/institutional power and you’re not trying to address this problem, you are doing a great disservice to science and young scientists.
This is a super interesting emergent phenomenon in LLMs that I think is actually quite profound.
The Waluigi Effect (mega-post) — LessWrong
Everyone carries a shadow, and the less it is embodied in the individual’s conscious life, the blacker and denser it is. — Carl Jung …
www.lesswrong.com
March 3, 2025 at 6:39 PM
This is a super interesting emergent phenomenon in LLMs that I think is actually quite profound.
Reposted by Alex P. Miller
Reposted by Alex P. Miller
link 📈🤖
Pareto optimal proxy metrics (Zito, Greaves, Soriano et al) North star metrics and online experimentation play a central role in how technology companies improve their products. In many practical settings, however, evaluating experiments based on the north star metric directly can be diff
Pareto optimal proxy metrics (Zito, Greaves, Soriano et al) North star metrics and online experimentation play a central role in how technology companies improve their products. In many practical settings, however, evaluating experiments based on the north star metric directly can be diff
February 20, 2025 at 5:00 PM
link 📈🤖
Pareto optimal proxy metrics (Zito, Greaves, Soriano et al) North star metrics and online experimentation play a central role in how technology companies improve their products. In many practical settings, however, evaluating experiments based on the north star metric directly can be diff
Pareto optimal proxy metrics (Zito, Greaves, Soriano et al) North star metrics and online experimentation play a central role in how technology companies improve their products. In many practical settings, however, evaluating experiments based on the north star metric directly can be diff
Good video on the evolution of nonchalance. I think with AI making the outputs of art easier to create, we will naturally start to place much more value on things only humans can do. One of those things is to exercise agency and show people that you actually care enough to try.
How Trying Became Cool Again
YouTube video by Nathan Zed
www.youtube.com
February 17, 2025 at 8:22 PM
Good video on the evolution of nonchalance. I think with AI making the outputs of art easier to create, we will naturally start to place much more value on things only humans can do. One of those things is to exercise agency and show people that you actually care enough to try.
Reposted by Alex P. Miller
If we have any hope of preventing the collapse of US scientific and research environment, we also have to rip off our blinders on the ways our institutions are weak and how they have failed so many.
Those weaknesses are points of leverage for people trying to tear the whole thing down.
Those weaknesses are points of leverage for people trying to tear the whole thing down.
February 16, 2025 at 5:38 PM
If we have any hope of preventing the collapse of US scientific and research environment, we also have to rip off our blinders on the ways our institutions are weak and how they have failed so many.
Those weaknesses are points of leverage for people trying to tear the whole thing down.
Those weaknesses are points of leverage for people trying to tear the whole thing down.
Reposted by Alex P. Miller
On the white liberal woman as the Omni-villain for the right and the far left
www.cartoonshateher.com/p/liberals-w...
www.cartoonshateher.com/p/liberals-w...
January 28, 2025 at 2:07 PM
On the white liberal woman as the Omni-villain for the right and the far left
www.cartoonshateher.com/p/liberals-w...
www.cartoonshateher.com/p/liberals-w...
Science is in an impending crisis of epistemology. Old, trust-based methods which many were happy to claim were sufficient will stand no chance in the future.
Academia is in trouble - finance scholars show how "AI systems can rapidly produce hundreds of seemingly coherent theoretical explanations for mined empirical results" papers.ssrn.com/sol3/papers.... GitHub of the 288 papers they "wrote" as a demo here github.com/velikov-miha...
AI-Powered (Finance) Scholarship
This paper describes a process for automatically generating academic finance papers using large language models (LLMs). It demonstrates the process' efficacy by
papers.ssrn.com
January 21, 2025 at 7:50 PM
Science is in an impending crisis of epistemology. Old, trust-based methods which many were happy to claim were sufficient will stand no chance in the future.
David Lynch was a master of capturing the inchoate darkness and beauty that exists outside of what we can articulate in words alone. He inspired me and countless others with his work.
RIP to a true artist, visionary, and virtuoso of vibes.
RIP to a true artist, visionary, and virtuoso of vibes.
David Lynch, Visionary Director of ‘Twin Peaks’ and ‘Blue Velvet,’ Dies at 78
Director David Lynch, who radicalized American film with with a dark, surrealistic artistic vision in films like 'Blue Velvet,' has died. He was 78.
variety.com
January 16, 2025 at 6:54 PM
David Lynch was a master of capturing the inchoate darkness and beauty that exists outside of what we can articulate in words alone. He inspired me and countless others with his work.
RIP to a true artist, visionary, and virtuoso of vibes.
RIP to a true artist, visionary, and virtuoso of vibes.