Daily Log Digest – Week 28, 2025
2025-07-12
AI Therapy
AI therapy bots fuel delusions and give dangerous advice, Stanford study finds - Ars Technica #ai #therapy
Given these contrasting findings, it's tempting to adopt either a good or bad perspective on the usefulness or efficacy of AI models in therapy; however, the study's authors call for nuance. Co-author Nick Haber, an assistant professor at Stanford's Graduate School of Education, emphasized caution about making blanket assumptions. "This isn't simply 'LLMs for therapy is bad,' but it's asking us to think critically about the role of LLMs in therapy," Haber told the Stanford Report, which publicizes the university's research. "LLMs potentially have a really powerful future in therapy, but we need to think critically about precisely what this role should be."
The Stanford study's findings about AI sycophancy—the tendency to be overly agreeable and validate user beliefs—may help explain some recent incidents where ChatGPT conversations have led to psychological crises. As Ars Technica reported in April, ChatGPT users often complain about the AI model's relentlessly positive tone and tendency to validate everything they say. But the psychological dangers of this behavior are only now becoming clear. The New York Times, Futurism, and 404 Media reported cases of users developing delusions after ChatGPT validated conspiracy theories, including one man who was told he should increase his ketamine intake to "escape" a simulation.
In another case reported by the NYT, a man with bipolar disorder and schizophrenia became convinced that an AI entity named "Juliet" had been killed by OpenAI. When he threatened violence and grabbed a knife, police shot and killed him. Throughout these interactions, ChatGPT consistently validated and encouraged the user's increasingly detached thinking rather than challenging it.
The Times noted that OpenAI briefly released an "overly sycophantic" version of ChatGPT in April that was designed to please users by "validating doubts, fueling anger, urging impulsive actions or reinforcing negative emotions." Although the company said it rolled back that particular update in April, reports of similar incidents have continued to occur.
Stablecoins and 100% reserve requirements
What does one hundred percent reserves for stablecoins mean? - Marginal REVOLUTION #crypto #stablecoin #reserves
The statute’s policy goal is to keep a payment‑stablecoin issuer from morphing into a fractional‑reserve bank or a trading house while still giving it enough freedom to:
- hold the specified reserve assets and manage their maturities;
- use overnight Treasuries repo markets for cash management (explicitly allowed);
- provide custody of customers’ coins or private keys.
Everything else—consumer lending, merchant acquiring, market‑making, proprietary trading, staking, you name it—would require prior approval and would be subject to additional capital/liquidity rules.
Why Your Brain Gets High on Uncertainty #neuroscience #brain #uncertainity
But, despite all this change, we’ve adjusted nicely to our new high-tech world. Why? Because we thrive on a challenge. We thrive on the uncertainty that comes with learning new things.
But why would our brains evolve to thrive on uncertainty? Shouldn't we prefer certainty, like knowing exactly where our next meal is coming from?
As I mentioned earlier, uncertainty was critical for our survival. Think about our ancestors who conquered new lands. The ones who were curious about what might be over that next mountain range?
So what can we do with this knowledge? Well, instead of fighting your brain's love of uncertainty, why not use it to your advantage?
- Want to learn something new? Frame it as a mystery to be solved.
- Need to exercise more? Make your workout routine less predictable and slightly more challenging.
- Trying to stay motivated at work? Gamify projects with elements of discovery and reward.
Like anything pleasurable, too much of a good thing can ruin it. Like too much candy for a nickel. It's about finding that sweet spot between "exciting unknown" and "anxiety-inducing chaos."
Meditation and Boredom
Find meditation really boring? You’re not the only one | Psyche Ideas #meditation #boring
In fact, what my colleagues and I call ‘spiritual boredom’ has a long tradition. Christian history contains numerous depictions of boredom: paintings of yawning congregants, people sleeping during sermons, and so on. In the Middle Ages, this phenomenon was recognised as a spiritual malaise called acedia (from Latin), characterised by listlessness and melancholy. Christians referred to it as the ‘demon of noontide’ – a concept described by St Thomas Aquinas as the ‘sorrow of the world’ and the ‘enemy of spiritual joy’.
Beyond these examples from Christian history, reports of boredom can be found in almost every spiritual practice. For instance, in Buddhist contexts, there are accounts of boredom during Asanha Bucha Day sermons. Similarly, some reports relating to mindfulness meditation describe experiences of ‘void’ – an emotional state combining boredom and psychological entropy.
Having said all that, I don’t believe boredom is just an obstacle – it could also be informative. From an evolutionary perspective, boredom exists to signal misalignment. It’s your brain’s way of saying: ‘This doesn’t suit you – change something.’ If you ever find yourself bored while meditating, praying or listening to a sermon, it might be helpful to ask yourself: ‘Am I over- or underchallenged?’ and ‘Does this practice (still) hold personal meaning for me?’
Postgres LISTEN/NOTIFY
Postgres LISTEN/NOTIFY does not scale | Hacker News #postgres #pubsub #queues
This is an interesting HN thread about the scalability limitations of LISTEN/NOTIFY. The blog post is worth reading. What caught my attention was this thread which had some interesting discussion
This is roughly the “transactional outbox” pattern—and an elegant use of it, since the only service invoked during the “publish” RPC is also the database, reducing distributed reliability concerns.
…of course, you need dedup/support for duplicate messages on the notify stream if you do this, but that’s table stakes in a lot of messaging scenarios anyway.
Yeah, but pub/sub systems already need to be robust to missed messages. And, sending the notify after the transaction succeeds usually accomplishes everything you really care about (no false positives).
Boosterism
Boosterism - by Rob Kurzban - Living Fossils #evo-psych #heirarchy #power
Boosting seems to have to do with cases in which an individual2 does something—I’m going to call it the Thing, with a capital letter—that that individual is either not allowed to do, by convention or rule, or is stereotypically not good at—according to the current cultural norms, or both.
…
Boosterism seems to be the feeling you get when someone does someThing stunning and brave that fits the scheme above.
Why?
Scholars such as Chris Boehm—see, for instance, his book Hierarchy in the Forest—have suggested that humans have a propensity to try to flatten hierarchies. As we have seen in posts about power, when there is one individual—or a group of a few individuals—who everyone else always backs, these few powerful people can do practically whatever they want, advancing their (fitness) interests at the expense of others’. Boehm suggests that humans naturally want to limit the power of the powerful. Certainly there is cross-cultural evidence of this preference, especially in the so-called collectivist cultures associated with Asia.
This resonates with boosterism, if imperfectly. The story about the marathon can be seen as part of eroding the power of men in society, reducing the extent to which it is an identity-focused regime, as I’ve called it. Generally, boosterism feels anti-hierarchy. So maybe boosterism is a leveling system, designed to support underdogs to prevent domination by the few, or the one. It’s probably often fitness-good to support the erosion of power of people or groups who can impose their will on you. Leveling is good for those who aren’t part of the elite.
As some researchers put it, “[a]lthough people prefer to associate with winners, there is also a strong desire to support the lovable loser or underdog.” It feels good to stand up and say, yes, I too support people doing that Thing.
But if everyone else has the same belief, well, that’s neither particularly stunning nor especially brave. When the battle is long over, and the moral arc has fully arced, boosterism changes. It still feels good—but it’s no longer subversive. It’s orthodoxy in the costume of rebellion. And like all such performances, it risks slipping into the theater of the absurd: applause lines for acts no longer forbidden, cheers for victories already won.
2025-07-13
What I could have learnt from René Girard #mimetic #culture
René Girard might have found metaphorical use for this. The French theorist’s great idea was that religion and culture grow out of what he called mimetic rivalry. Human beings, uniquely, choose the objects of their desire largely on the basis of what other people desire. “There is nothing, or next to nothing, in human behaviour that is not learned, and all learning is based on imitation,” he writes. But while mimesis helps us learn, it also leads to escalating competition, and ultimately violence. Religion evolved as a means for containing rivalry by projecting communal violence on to an arbitrarily chosen sacrificial victim, the scapegoat.
As always happens when an intellectual becomes popular, distortions have followed. The main problem, though, is not misinterpretation. It’s omission. What is often left out of discussions of Girard is the most challenging part of his theory, about how we break the cycle. Here he turns to one of the firmest messages of the gospels: the injunction to love our enemies. Girard knew, as we all know, that renunciation and mercy are almost impossibly hard, and quite alien to human culture. Yet he argues that it is the moments when the mimetic crisis has reached a hysterical crescendo, when “the vanity and stupidity of violence have never been more obvious”, that it is possible to see our enemies in a new way. Might we not be living in such a moment right now?
Extraction vs Creation
From Dollar Dominance to the Slop Machine - by kyla scanlon #economics
The US has become an extraction economy.
- We extract value from our existing position through dollar dominance, military supremacy, and technological leadership and now are choosing to tear down the foundations that created that position in the first place.
- We extract attention through spectacle without creating the trust that makes spectacle meaningful.
- We extract wealth from our own institutions without replenishing the capacity that generated that wealth.
- The UFC image captures this well - it takes the symbolic power of American institutions and converts it into entertainment value, with no consideration for what that conversion costs us in terms of credibility or coherence.
China, meanwhile, has become a creation economy1.
- They're building electrical generation capacity, training engineers, developing industrial policy that spans decades.
- They're creating an “electrostate” with an economy driven by the technologies that will determine 21st-century competitive advantage.
Immigration Policy of the Danish left
Denmark’s left defied the consensus on migration. Has it worked? #denmark #immigration
These are uncomfortable facts, so much so that to point them out is to invite the disgust of European polite society. Whether in France, Germany, Italy or Sweden, parties of the hard right have surged as they—and often only they, alas—persuaded voters that they grasped the costs of mass migration. But the National Rally of Marine Le Pen in France and Giorgia Meloni’s post-fascist Brothers of Italy have an unexpected ally: Denmark’s Social Democrats, led by the prime minister, Mette Frederiksen. The very same party that helped shape the Scandinavian kingdom’s cradle-to-grave welfare system has for the past decade copy-pasted the ideas of populists at the other end of the political spectrum. Denmark is a generally well-run place, its social and economic policies often held up for other Europeans to emulate. Will harsh migration rhetoric be the next “Danish model” to go continental?
The Danish left’s case for toughness is that migration’s costs fall overwhelmingly on the poor. Yes, having Turks, Poles or Syrians settle outside Copenhagen is great for the well-off, who need nannies and plumbers, and for businesses seeking cheap labour. But what about lower-class Danes in distant suburbs whose children must study alongside new arrivals who don’t speak the language, or whose cultures’ religious and gender norms seem backward in Denmark? Adding too many newcomers, the argument goes—especially those with “different values”, code for Muslims—challenges the cohesion that underpins the welfare state.
The upshot of the left’s hardline turn on migration has been to neutralise the hard right. Once all but extinct, it is still only fifth in the polls these days, far from its scores in the rest of Europe. For good reason, some might argue: why should voters plump for xenophobes when centrists will deliver much the same policies without the stigma? Either way, that has allowed Ms Frederiksen to deliver lots of progressive policies, such as earlier retirement for blue-collar workers, as well as unflinching support for Ukraine. The 47-year-old is one of few social-democratic leaders left in office in Europe, and is expected to continue past elections next year.
2025-07-15
Coffee and aging
Coffee May Promote Longer, Healthier Living - Bloomberg #caffeine #anti-aging
If you're like me, you climb out of bed each morning feeling like a zombie -- until you slug back that first cup of coffee.
Turns out that morning jolt may benefit more than just energy levels. It could help slow down the aging process of the body's cells, potentially helping to fend off ailments including cancer and neurodegeneration.
Caffeine flips a biological switch in our bodies called AMPK, which monitors our cells' energy levels and, when they're low, tells them to slow down their growth processes and instead focus on repairing damage, according to a paper recently published in the journal Microbial Cell.
In doing so, caffeine inhibits the cellular growth regulator TOR, explains Babis Rallis, the paper’s senior author and a reader in genetics, genomics and fundamental cell biology at Queen Mary University of London. TOR is highly active when we're embryos and fast-growing kids, helping us develop into adults. Once we're older, it will contribute to our body’s ability to, say, renew skin, grow hair and heal wounds.
TOR, however, is "pro-aging." When it's too active, it's implicated in problems including metabolic disorders, neurodegeneration, inflammation and cancer, according to Rallis.
By studying caffeine's effect on cell growth, Rallis is hoping to get a better handle on some of the factors that promote longevity, a field known as biogerontology. That could inform future research into how we can trigger these virtuous cellular effects through diet, lifestyle and new medicines to achieve healthy aging, he said in an interview.
"We're not saying that you have to take hundreds of pills, like we have seen in the news by certain billionaires," he said. "We mostly try to uncover biological mechanisms and understand how you can then change your habits."
2025-07-16
Google Finds a Crack in Amazon’s Cloud Dominance
Google Finds a Crack in Amazon’s Cloud Dominance — The Information #gcp #google #aws #amazon #ai #startups
AWS generates more than twice as much revenue as Google Cloud and has long dominated the market for selling cloud services to startups. But the Dia episode and other examples show how Google has become competitive in attracting AI startups to its cloud, thanks to Gemini and other capabilities AWS doesn’t have.
AWS’ struggle to develop a strong AI model of its own has fueled a perception that it is trailing Google in developing cutting-edge AI.
That’s a big turnaround after Google’s earlier struggles with previous versions of Gemini and startups’ widespread complaints about the difficulty of setting up Google Cloud accounts for AI computing.
Google Cloud has even landed business from two high-profile AI startups its own AI teams compete with: Safe Superintelligence Inc., led by former OpenAI Chief Scientist Ilya Sutskever; and Thinking Machines Lab, helmed by former OpenAI Chief Technology Officer Mira Murati. (AWS may have wanted Murati’s business, too: Amazon CEO Andy Jassy met with her in San Francisco earlier this year.)
Google Cloud also recently won business from an even fiercer rival, OpenAI, which has been a major customer of Microsoft cloud servers but has been branching out to other providers.
Aruna Roy on Jane Austen
Social activist Aruna Roy on Jane Austen’s enduring appeal - The Hindu
As we mark the 250th anniversary of Austen’s birth, I am reminded of British-American poet W.H. Auden’s remark about her in his ‘Letter to Lord Byron’.
…It makes me most uncomfortable to see
An English spinster of the middle-class
Describe the amorous effects of ‘brass’,
Reveal so frankly and with such sobriety
The economic basis of society…
Raising Children
Learning to Parent in Community – SAPIENS #parenting #caregiving #culture #anthropology
Bebuna, a woman in her 60s, sits in front of her hut breastfeeding her granddaughter. I had never seen an older woman nursing and—even as an evolutionary anthropologist—didn’t realize it was biologically possible.
“Are you producing milk?” I asked.
Bebuna squeezed her breast, and white droplets appeared.
It turns out that lactating people can produce milk indefinitely, as long as they continue breastfeeding baby after baby. Bebuna has been doing just that for decades, as a midwife and caregiver to many children in her community.
Bebuna is a member of the BaYaka, a collective name for several forest-living forager groups west of the Congo River. [1] Her community lives in the northern Republic of Congo and speaks the Mbendjele language. In 2013, I began visiting Mbendjele BaYaka camps to research how people learn from others in a hunter-gatherer society.
There are countless ways to raise children, shaped by cultural traditions or, in more individualistic societies, by personal choices among various parenting philosophies. All approaches bring benefits and challenges. But community-oriented cultures like the BaYaka get at least one thing right: Parenting should not be learned in isolation or all at once—it is a lifelong process embedded in daily life long before one has a child and long after.
Unlike me, the BaYaka learn to parent before they become parents.
Among both babies under 1.5 years and children aged 1.5–4, around 40 percent of their close care, including holding and physical contact, was provided by “allomothers”—caregivers other than the biological mother. On average, each child had 14 people within arm’s reach throughout the day.
Mothers responded to just under half of all crying bouts. Allomothers soothed the rest—over 40 percent on their own, the remainder alongside the mother. Soothing often meant drumming on the child’s back or yodeling to gently calm the child.
And who were these allomothers? Mostly, other children. These young helpers were more involved, collectively, than fathers or grandmothers.
Parenting is never perfect. Cultures raise children differently, shaping adults valued by their own standards. But, based on my experiences, one truth emerges: Learning to care for others should start long before having a baby.
Two years into parenthood, I have more questions than answers. Growing up in Turkey, studying across Europe and Canada, and working in the U.K., I had never held a newborn until I had Eren. I spent my childhood and early adulthood learning subjects like math, physics, and literature—what my societies valued most. My first months of motherhood were emotionally overwhelming because of the steep learning curve I had to scale.
I wish, like the BaYaka, my parenting lessons had followed a gentler slope, stretched across my lifespan. The same could be said about other essential life skills like growing food, caring for our elders, and dealing with death.
For those of us living in individualistic societies, what happened that people stopped caring to learn life’s basics? Take a note from the BaYaka and other community-oriented cultures: Bring these lessons back into learning journeys.
What’s Happening to Reading?
What’s Happening to Reading? | The New Yorker #reading #ai
What will happen to reading culture as reading becomes automated? Suppose we’re headed toward a future in which text is seen as fluid, fungible, refractable, abstractable. In this future, people will often read by asking for a text to be made shorter and more to-the-point, or to be changed into something different, like a podcast or multi-text report. It will be easy to get the gist of a piece of writing, to feel as if you know it, and so any decision to encounter the text itself will involve a positive acceptance of work. Some writers will respond by trying to beguile human readers through force of personality; others will simply assume that they’re “writing for the A.I.s.” Perhaps new stylistic approaches will aim to repel automated reading, establishing zones of reading for humans only. The people who actually read “originals” will be rare, and they’ll have insights others lack, and enjoy experiences others forgo—but the era in which being “well-read” is a proxy for being educated or intelligent will largely be over. It will be difficult to separate the deep readers from the superficial ones; perhaps, if A.I.-assisted reading proves useful enough, those terms won’t necessarily apply. Text may get treated like a transitional medium, a temporary resting place for ideas. A piece of writing, which today is often seen as an end point, a culmination, a finished unit of effort, may, for better and worse, be experienced as a stepping stone to something else.