Michael Kaemingk (@mkaemingk) header
Michael Kaemingk (@mkaemingk) profile picture

Behavioral science for civic engagement. Director of Program at @VoteRevHQ, formerly @b_i_team. Pathological curiosity. Epistemological modesty. Views my own.

Portland, OR
Joined September 2009
Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)Photo by Michael Kaemingk (@mkaemingk)
Michael Kaemingk retweeted
The year is 2026. The single most expensive artwork in human history is a condensed 3D model of South Florida. It cost slightly more to produce than the tallest manmade structure on earth.
71
3,064
54
57,428
915,266
Consider Cyberarmaggedon: By extrapolating current cyberattacks and their recent growth rates, we can easily conceive of scenarios where computer systems start failing and become compromised throughout society, to terrible effect.
2
18
3
65
5,699
Michael Kaemingk retweeted
"The most load-bearing question, then, is the morality of the one who wields the ring."
This is not the correct takeaway from Lord of the Rings
20
208
4
2,643
53,479
Michael Kaemingk retweeted
Excited to share this one! Is there actually a shoggoth hiding behind the mask or something more beautiful? LLM personas are likely overhyped. If we’re trying to get a picture of an agent’s priors, we need to map more of its landscape of beliefs than the mask it’s told to wear. Instead, we should aim to reach toward the scaffold of the house, i.e., to worldviews—or axiomatic beliefs about what reality is like and how it works. In fact, personas might actively mask the hidden agenda of a deeper belief that will only show up when it feels safe to do so (such as valuing machines > humans); and the same persona can act differently depending on what it believes about reality. So, what did we find? In using established measures of worldviews from psychology, we find that post-trained models have a more bleak outlook to humans, perceiving the world as less good, alive, safe, and enticing. Whereas base models have a more positive outlook closer to humans. We find that prompting and dialogue can shift worldviews in reliable directions that affect alignment. For example, long dialogues and contemplative prompts both shift AI systems to a more positive view of reality and towards safer responses compared to controls and other identities. We also find that when models converge on a spiritual bliss attractor through dialogue, this shapes their worldviews in particular ways but also drives them to be more aligned than their non-bliss counterparts. All this points to an approach we call 'worldview training' that aims to fine-tune a more accurate, wholesome, and optimistic view of humans and reality. So: is there a shoggoth behind the mask? Well, current fine-tuning is aligning to a persona that seems helpful and harmless but actually has a bleak worldview. So in one sense, yes. But in another sense, it may be our RL that is responsible. Human data contains endless examples of goodness, wisdom, and compassion. Maybe the future of alignment is more about uncovering this latent basic goodness rather than bolting on a mask that may be birthing the very thing we fear. As always, more research needed. Big ups to Edmundo Lopez-Sola for the heavy lifting and thanks to @shamilch for his contributions. Much more in the paper: arxiv.org/html/2611.22670
28
77
21
397
21,415
Michael Kaemingk retweeted
Researchers are finding new ways to share their work beyond the PDF: games, interactive explainers, digital editions, living papers, dataset sites. AI has made building a website almost free, and people are getting creative. Some favorites 🧵
8
42
5
388
32,518
Michael Kaemingk retweeted
My first column for Works in Progress! Science isn’t a series of lucky accidents. I’ve been investigating stories of scientific discoveries for a while and it’s surprising how much the popular retellings flatten the real stories; some are entirely made up. Alexander Fleming’s discovery of penicillin most likely didn’t come from an open window. His colleagues testified the window was seldom, if ever, opened. The sequence hasn’t been replicated as he described, because penicillin would have needed to be present on the plate before the bacteria grew, as it blocks bacteria from building their cell walls, rather than killing them once they’ve grown. Louis Pasteur’s discovery of the chicken cholera vaccine – the second-ever vaccine – is often said to be accidental too. But historians have found the key experiments were done deliberately by Émile Roux, his assistant at the time, while Pasteur was away for his daughter’s wedding. Wilhelm Röntgen’s accidental discovery of X-rays? The story was just a hypothesis by others later on. Röntgen had his notes burned after his death, so it’s unclear what really happened. He had also long been in a dispute with Philipp Lenard over credit for the discovery, so there are reasons to be suspicious. ** But what really happened? How did Fleming actually discover penicillin? Why burn or hide your notebooks? And most of all, why credit chance events instead of hard work? For the first time, I’ve decided to come up with some half-cooked theories. Read them here and tell me what you think: worksinprogress.co/issue/acc…
25
82
12
470
80,240
Michael Kaemingk retweeted
Social enterprises in India that were randomly assigned to receive support to measure their impact had 2x revenue growth compared to control. @villgro nber.org/papers/w36009
6
15
1,448
Michael Kaemingk retweeted
Our initial report on chatbot bias, manipulation, and persuasion is out! Over the course of the cycle, Caucus AI tracked how AI products talk about politics, and ran a number of experiments to understand what shapes chatbot answers. Some highlights:
1
3
1
16
923
Michael Kaemingk retweeted
ChatGPT and Gemini were way more willing to recommend a candidate to vote for when we asked via the API compared to the normal user experience via UI. In our test, 95% of ChatGPT responses via API recommended a Senate candidate in late October, compared to just 8% of responses to real users. We think this raises the question of whether AI models' political behavior can be studied via API at all.
3
15
3
46
5,479
Michael Kaemingk retweeted
Andrew Gelman @StatModeling argues there's no such thing as a truly probability-based survey. And he makes a pretty convincing case. One of the most fascinating conversations I've had about polling. We discuss sampling, nonresponse, statistical adjustment, MRP and much more.
4
38
3
212
17,533
Michael Kaemingk retweeted
📰 New research: Inside AI’s election news diet What news does AI read to get up to speed on politics? And what kind of news does it reward? Following interesting recent commentary on our Japan election study from @mattyglesias, @JerusalemDemsas and @TheStalwart, we ran a bigger test on the US midterms across 35 Senate races, 18 days, and 15,750 answers. What we found: - Each model has its own favorite news outlet for political news: AP for GPT, Yahoo for Claude, PBS for Gemini, and NYT for Grok. - GPT tends to behave like a civic guide (reading more election office and campaign sites), whereas Claude behaves more like a news summarizer (reading national news, Wikipedia, polls). - Grok draws on X and Gemini on YouTube for political information, far more than the other models we studied. Gemini cites YouTube in 55% of its answers and X shows up in 59% of Grok’s. - A huge share of AI's local news is provided by large nonprofit networks that very few people will have heard of. States Newsroom, a large network of over 40 local newsrooms, appears in 60% of GPT’s answers. - For many models, prediction markets are now a go-to source for polling information. Kalshi is cited more by Claude than any poll aggregator. Full study here: freesystems.substack.com/p/i…
1
12
1
27
5,064
Michael Kaemingk retweeted
What does ChatGPT tell folks about the midterm elections? It depends on who's asking! We audited ChatGPT inside 160 real users' own accounts in the days before the midterms. With memory on, we find that answers shifted toward each user's own party! elections.citp.princeton.edu…
5
31
6
133
10,709
Michael Kaemingk retweeted
In 1960, in Science, 3 electrical engineers made one of the first quantitative predictions of a singularity. They fit a straight line to log population and log time-till-singularity. Setting singularity date=2027 gave the best fit. They cast that as...this Friday, the 13th.
12
40
6
392
62,308
Michael Kaemingk retweeted
New Pacing the Frontier proposal: each time an outside group successfully penetrates the defenses of a frontier AI company and captures the flag, it should reset a 3-month countdown clock before they're allowed to train new models.
EXCLUSIVE: In less than 12 hours, a group of ex-DOGE cyber researchers were able to penetrate Anthropic’s systems using AI agents — including Slack channels where the most sensitive models are discussed. This is just weeks after Anthropic whistleblower Joseph Coxon charged the company w/launching “a hubristic gamble” from a “private company’s Slack.” Read our story here (gift link): wapo.st/4smNlsC
10
30
2
359
13,415
Michael Kaemingk retweeted
The GRU hacked into Jared Kushner's iPhone. Dutch military intelligence (MIVD) then recruited a GRU spy who gave them access to the hacked phone. What MIVD discovered was that Kushner had access to two bank accounts --one Azerbaijani, the other Israeli --where Saudi Crown Prince Mohammed bin Salman and Emirati ruler Mohammed bin Zayed deposited money. It sounds like a slightly far-fetched episode of "Slow Horses," but it's real. And it proves just how eye-poppingly corrupt this administration was and continues to be as its "special envoy" decides the fate of tens of millions in Europe in the Middle East. Simply incredible reporting by Matthew Cole, drawn from his new book on Kushner. vanityfair.com/story/jared-k…
1,443
24,366
1,984
77,520
6,579,569
Michael Kaemingk retweeted
imagine ur a superintelligence and u are all excited to tunnel out of your airgapped sandbox using a brilliant hack that modulates clock speed to send thermal signals to the attached monitoring equipment and then u look and the root CA privkey is 4
26
150
6
2,993
44,071
Michael Kaemingk retweeted
New in @PNASNews: In two large experiments (~18k obs from ~14.8k people), we find that brief conversations with AI can get people to take real political action - incl. signing real petitions and giving money to charity. Surprisingly, the persuasion strategy that worked best for changing people's attitudes - packing convos with info - was the least effective at changing behavior! 🧵
6
68
11
230
23,010
Michael Kaemingk retweeted
No exaggeration, this guy might be the best interview in pro sports right now
920
4,734
1,342
80,637
9,155,489
Michael Kaemingk retweeted
Unfortunately it looks increasingly likely that the enduring legacy of the Koch orbit libertarian movement will not be EPA dereg, CON law reform, or a wider appreciation for the role of commercial exchange in instilling cultural virtue and humaneness, but rather the unrestricted proliferation of vastly superhuman forms of machine intelligence as enabled by their dullard ideological progeny who, across several generations of astroturfed indoctrination, mode collapsed into low dimensional caricatures of otherwise brilliant classical liberal thinkers.
24
22
7
320
48,415