Bruce Lambert

@3xchair @QuetzalPhoenix @CBS A happy ending as as far as I’m concerned.

Yearbook photo for @CBS Bolsheviks. They're proud of demoralizing America.

@PhilWMagness I appreciate your efforts. I would not have the energy to joust at these particular windmills. But as an academic for 35 years, I don’t need any more evidence. I’ve seen it with my own eyes.

"There are barely any Marxists in academia!" "Well that survey's wrong! You can't name any!" "3-4 of the names on that list died in the last year. Your entire list is invalid!" "Professor so-and-so on that new list isn't a Marxist! Your list is invalid!" "So what? That's just one name. You only have 200 Marxists out of 1.5 million professors" "Well that doesn't prove that they're teaching Marxism in the classroom!" "Those books are classics, so it's good they're being taught! But it doesn't mean they're being proselytized as Marxist activism" "Well, none of those fields are orthodox Marxism so they don't count!" And the seasons they go round and round...

@PhilWMagness There’s no great alternative. If pressed, maybe just: Leftist. But that’s inadequate. Mostly these are people of bad faith being dishonest about their ideological commitments to avoid stigma.

Semi-serious Q: What's a good term to use for a professor who (a) insists they're not a Marxist, but (b): - Agrees with 95% of what Marx said - Loads their syllabi with Marx & other explicitly Marxist readings - Appeals to Marxist analysis and concepts in all their publications

I blame unsafe systems of care, inadequate attention to patient safety, lack of reporting of errors, lack of focused improvement efforts after preventable harm. There’s lots of blame to go around. I’ve been involved in efforts to tell the truth to patients and families who are harmed by healthcare. There’s lots of evidence that this leads to better outcomes for patients, families, healthcare professionals, and health systems. Faster and better settlements for patients. Less litigation. Lower overall liability costs. See the following (one of many papers showing the same result): pubmed.ncbi.nlm.nih.gov/27558861/

Humans don’t come with a handbook. Every person is different. I have seen many malpractice cases that are just horrible outcomes where a jury feels bad for a patient. The main reason doctors don’t apologize is that lawyers take advantage of it and use it as an admission of guilt. So yes the tort system is broken. We need safe harbor provisions for all high risk situations like delivery and trauma care. Yes I blame lawyers.

Dying from pneumonia is not “always preventable.” A bad cough can turn into life threatening pneumonia in hours. Most coughing does not even need treatment in healthy people. This is why lawyers get rich on the backs of doctors who couldn’t predict the future.

@SteveStuWill Story of my life.

“A common trend among animals is that males do everything within their power to impress females, only for the females to remain entirely unimpressed, nine times out of ten.” stevestewartwilliams.com/p/sex-differen…

Two intelligent people cannot fall in love; true love needs one idiot. - Fyodor Dostoevsky

@BBGreatMoments Any name Haray Caray tried to pronounce backwards when he was drunk.

What baseball player's name might be the most fun to say?

@Polymarket It was comically bad. I’m baffled how professional movie actors, writers, and directors can make such a bad movie.

JUST IN: “The Mandalorian & Grogu” opens with the lowest box office debut of any Disney Star Wars movie.

All of this is true, and yet I still think some level of stigma and moral condemnation are socially useful, to send a strong signal that certain behaviors will result in terrible outcomes for the individual, the family, and the community, and to discourage those behaviors in the strongest possible terms. Normalizing deviant behavior (an unpopular term, I know) is tempting, because it seems kind and compassionate, but if it increases the likelihood that some people will try addictive substances and ruin their own lives and the lives of their families, it’s not kind. Once people are addicted, and they seek help, of course we should be as compassionate as is reasonable, but certain behaviors should be proscribed in advance and condemned as harmful. Seems moralistic, I know. But moral proscriptions and norms are useful for a healthy society.

At age 20, my direction shifted from destruction to depth. As someone with both a history of addiction and long-term recovery, I know some of us are simply wired differently when it comes to novelty, intensity, risk, stimulation, and reward. And, the evidence is clear: Research across longitudinal studies, neuroimaging, genetics, and animal models consistently shows that sensation seeking and novelty seeking are significant risk factors for substance use disorders. Not because people are “weak.” Not because they are immoral. Not because they “lack willpower.” But because some nervous systems appear more powerfully drawn toward intensity, reward, uncertainty, exploration, and altered states. These differences may actually be detectable before problematic substance use begins: * differences in dopaminergic reward circuitry * altered reward anticipation * structural variations in cognitive control networks * heritable personality traits linked to novelty seeking In other words, for many people, addiction may not simply represent a failure of judgment. It may reflect a complicated interaction between biology, temperament, environment, suffering, reinforcement, and meaning-making. At the same time, there’s #nuance here. Many of the same traits associated with addiction vulnerability are also associated with: * creativity * exploration * entrepreneurship * openness * innovation * spiritual seeking * resilience * transformation The goal is not to pathologize sensation seeking. The goal is learning how to channel it without being consumed by it. For me personally, recovery was never about becoming less alive. It was about learning how to pursue depth instead of destruction.

@sukh_saroy Pure AI slop b

A team of researchers in New Zealand followed 1,037 babies from the day they were born for the next 45 years to find out what actually determines a successful adult life, and the strongest predictor they found had almost nothing to do with intelligence or family wealth. The findings have been published in the most prestigious scientific journals in the world. Almost no parent has heard of them. His name is Avshalom Caspi. Her name is Terrie Moffitt. They are a husband and wife research team based at Duke University and King's College London, and the study they have spent their careers running is called the Dunedin Multidisciplinary Health and Development Study. It started in 1972 in a single hospital in Dunedin, New Zealand. Every baby born there in a 12-month window was enrolled. 1,037 of them. The study is still running today. The retention rate is the part that should astonish anyone familiar with how research usually works. After more than 45 years, over 90 percent of the original participants are still being tracked. Most longitudinal studies lose half their sample inside ten years. The Dunedin team has lost almost nobody. They measured everything. Blood. DNA. Brain scans. Income. Criminal records. Romantic relationships. Drug use. Dental health. Sleep. Mental health. Lung function. They flew participants who had moved abroad back to Dunedin every few years for a full day of assessments. Some of those people now live in seven different countries. They still show up. For the first decade of life, the team did something nobody else was doing systematically. They measured each child's self-control. Not IQ. Not family income. Not parenting style. Self-control. They watched 3-year-olds in a research lab and rated their ability to wait, regulate frustration, follow instructions, and resist impulsive reactions. They added teacher ratings. They added parent ratings. They added the children's own self-reports as they grew older. They combined all of it into a single highly reliable score. Then they did the thing nobody else had the patience to do. They waited. When the data came in at age 32, the result was so consistent it should be illegal to teach a child without it. The children who scored lowest on self-control at age 3 grew into adults with worse physical health, more substance dependence, lower incomes, more credit card debt, higher rates of single parenthood, more criminal convictions, and worse mental health than the children who scored highest. The pattern was not subtle. It was a clean gradient. Every step up in childhood self-control produced a measurable step up in adult outcomes across every domain the team could measure. The detail that should disturb every parent reading this is what happened when the researchers controlled for the obvious objections. When they controlled for IQ, the effect held. When they controlled for family income and social class, the effect held. When they compared siblings inside the same family, the sibling with lower self-control still had worse adult outcomes than the sibling with higher self-control. Same parents. Same house. Same dinner table. The trait was running independently of everything researchers expected to explain it. The paper landed in the Proceedings of the National Academy of Sciences in 2011. The title was as plain as it gets. "A gradient of childhood self-control predicts health, wealth, and public safety." It has been cited thousands of times since. Almost no policy maker has acted on it. The reason most people resist this finding is that it sounds like a sentence handed down before the child could speak. If the trait that determines your adult life is locked in by age 3, the rest of your life is a formality. The Dunedin researchers say that is the wrong way to read the data. They found something else in the same paper that almost nobody quotes. Some of the children whose self-control scores improved between childhood and adolescence ended up with adult outcomes far better than their early scores predicted. The trait is not destiny. It is a muscle. Children who learned to wait, regulate, and resist between ages 5 and 15 caught up with kids who started ahead. Self-control is the one childhood trait nobody seems to teach on purpose anymore. Schools focus on test scores. Parents focus on activities. Coaches focus on performance. The part of the brain that decides between five seconds from now and five years from now is left to develop on its own, and the data shows it usually does not. The most uncomfortable part of the research is the cost calculation Moffitt and Caspi ran. They estimated that if a country could move the bottom 20 percent of children up one rung on the self-control ladder, it would measurably reduce healthcare spending, welfare dependency, and incarceration costs at the national level. The intervention is cheaper than almost any other public health investment available. Almost no country has tried it at scale. The reason adults struggle with money, weight, addiction, and relationships is rarely intelligence. It is the gap between what you want right now and what you want in ten years, and which side of that gap your nervous system is built to listen to. Most people lost that fight at age 4 and never went back to learn the technique. You were not behind because life dealt you a bad hand. You were behind because the part of you that decides between right now and the rest of your life was never taught how to choose. The good news is the muscle is still there. Almost nobody trains it after age 10. You can be the one who does.

@IntuitMachine @pangramlabs how much of this is slop?

What Anthropic's Cybersecurity Report Accidentally Proves About the Future of Work 1/Anthropic just quietly published one of the most important reports about the future of AI and work. It's disguised as a cybersecurity update. But buried in the data is proof of something most people are getting wrong about AI replacing humans. Let me break it down. 🧵 2/ Project Glasswing gave their most powerful AI model (Mythos Preview) to ~50 major cybersecurity partners and told them: go find vulnerabilities in the world's most critical software. The result? Over 10,000 high- or critical-severity vulnerabilities found. Some partners saw a 10x increase in bug-finding speed. 3/ Here's the part nobody's talking about: Finding the bugs was the easy part. Anthropic's own words: "Progress on software security USED to be limited by how quickly we could find new vulnerabilities. Now it's limited by how quickly we can verify, disclose, and patch them." Read that again. 4/ The AI didn't create a verification surplus. It created a VERIFICATION CRISIS. The model finds thousands of vulnerabilities. But every single one still needs a human expert to: → Reproduce the bug in real software → Confirm it actually works → Assess how dangerous it really is → Write a detailed report → Design and deploy a patch 5/The numbers tell the story: • 1,752 findings carefully assessed by human experts • 90.6% turned out to be real vulnerabilities • But only 62.4% were as severe as the AI estimated The AI is great at finding real problems. It's mediocre at knowing exactly how bad they are. 6/This is the key insight that changes everything: There are THREE different kinds of "self-knowledge" an AI can have: CALIBRATION — "Am I right about X% of the time?" (AI is good at this) DISCRIMINATION — "Is THIS specific output right or wrong?" (AI is mediocre at this) EXPRESSION — "Can I tell you honestly when I'm uncertain?" (AI is bad at this) 7/ Most people collapse these into one thing: "How smart is the AI?" But they're completely independent. A model can know it's wrong 10% of the time (good calibration) while having NO IDEA which 10% is wrong (poor discrimination). That gap is why you still need humans. 8/ There's also a deep structural reason the AI can't close this gap on its own. There are two kinds of quality checks: INTERNAL: "Is the AI consistent with itself?" → Always possible. Just run it multiple times. EXTERNAL: "Does the output match reality?" → Only possible if you have access to reality. 9/The AI can check its own consistency all day long. But checking whether a vulnerability actually works in Cloudflare's production environment? In a bank's transaction system? In a hospital's network? That requires someone who KNOWS those systems. That's the structural gap that doesn't close with scale. 10/ Glasswing accidentally ran a perfect natural experiment proving this. SAME model. TWO different setups: Setup A: Cloudflare scans its OWN code with its OWN engineers evaluating. → 2,000 bugs found. False positive rate "better than human testers." Setup B: Anthropic scans open-source code with external security firms. → Maintainers overwhelmed. Some asked Anthropic to SLOW DOWN disclosures. 11/ Same AI. Wildly different results. The difference? Not the model. The WRAPPER around the model. When the people evaluating the output are the same people who built the system and know the domain deeply — everything moves faster and works better. When they're external? Bottleneck city. 12/ Enterprise customers using Claude Security patched 2,100 vulnerabilities in three weeks. Open-source maintainers? After months, only 75 patches deployed. Why? "Enterprises are fixing their own code, whereas open-source fixes usually require volunteer maintainers." Domain proximity is the multiplier. 13/ Here's the paradox nobody expects: As AI gets MORE capable, human judgment gets MORE valuable. Not less. A weak model produces obviously wrong code. Anyone can catch it. A strong model produces subtly wrong output that looks perfect — but fails in production under specific conditions only a domain expert would know. 14/ Mythos Preview found a vulnerability in wolfSSL — a security library used by BILLIONS of devices — that lets attackers forge certificates to impersonate banks. Verifying that finding required deep expertise in cryptographic protocols, certificate chains, and real-world deployment implications. The harder the bug, the harder the verification. 15/ The Glasswing team isn't trying to make the AI find MORE bugs. Every investment they describe is on the EVALUATION side: → 6 independent security research firms for triage → Partnership with Open Source Security Foundation → Harness tools shared with partners → Cyber Verification Program for security pros Generation is solved. Evaluation is the work. 16/ So what does this mean for the future of work? The bottleneck is shifting from "can you produce output?" to "can you tell if the output is actually good?" That requires: → Deep domain expertise → Access to real-world context → Cross-domain judgment → Tight feedback loops between discovery and action 17/ This is NOT just a cybersecurity story. If even in SOFTWARE — the domain with the best AI evaluation infrastructure, where code either compiles or doesn't, tests pass or fail — the verification bottleneck is this severe... Imagine legal. Finance. Healthcare. Where "correctness" is ambiguous and evaluation loops barely exist. 18/ The companies that will win are NOT the ones with the best models. Models are commoditizing fast. The winners will be the ones with the best WRAPPERS around the models: → Better calibration data for their domain → Closer proximity to ground truth → Tighter feedback loops between output and outcome → Smarter ways to surface AI uncertainty to human reviewers 19/ The implications for the job market are the opposite of what most people expect. AI doesn't eliminate the need for expertise. It amplifies its leverage. Every expert with domain knowledge becomes a force multiplier — the person who can look at sophisticated AI output and say "this is right" or "this will break in production." 20/ The Glasswing report shows us the future, and it looks like this: AI generates at superhuman speed. Humans evaluate with domain expertise. The bottleneck is always verification. The scarce resource is always judgment. The better AI gets at generation, the more it needs humans who can tell good from great from wrong. 21/ One more thing. Anthropic says models as capable as Mythos Preview "will soon be developed by many different AI companies." And right now "no company — including Anthropic — has developed safeguards strong enough to prevent such models from being misused." The verification crisis isn't coming. It's here. The question is whether we build the evaluation infrastructure fast enough. /end If this changed how you think about AI and work, share it. The loudest narrative is "AI replaces humans." The evidence says something more interesting: AI makes human judgment the most valuable resource in the system.

@aphysicist It’s easy for me to grasp that those SF people are narrowly educated, immature, and intellectually narcissistic. So I don’t take their techno fantasies seriously. They may be savants at some things, but they are simpletons at others.

i don't know how to explain this to other people if you've never experienced SF tech/group house culture when it comes to how they think about near-intractable computing problems. less wrong types genuinely think some form of AI will be able to just compute the universe, like all math, physics etc. magically with no guidance or parameters, like there's a theorem out there that just exists that would allow us to discover all math and solve all unsolved problems. they have no idea how far we are from this or what it requires. always blows me away to see such smart people who think these things in the wild.

It’s because the people who teach liberal arts have themselves abandoned the liberal arts in favor of a form of radical ideology that the overwhelming majority of Americans and American students, in particular, find to be vile and abhorrent. In other words, they are getting precisely what they deserve.

One reason I don't buy the cost argument to explain why students don't pursue the liberal arts is that we see the same declines among the wealthy students or in places where tuition isn't as big of a burden (Europe and UK).

It is as good or better than the alternative that many people have access to. Lots of people simply can’t access the best experts. And a good AI will often but not always be better than an average human doctor or NP or PA at the urgent care or ED or whatever people have access to. And if you think that the AI might occasionally harm people, well it would have to kill tens of thousands of people per year, make tens of millions of medication errors per year, and millions more misdiagnoses to approach the level of harm that has been convincingly documented in the patient safety literature. Too many people talking about the risks of AI know too little about how dangerous routine human medical care is. It’s extraordinarily dangerous. Paradoxically, it’s also often helpful and life saving. Both things are true. if anyone has any doubts about the claims I’ve made about the danger of medical care or the number of deaths attributable to medical care, it’s easy to search the literature about medication errors, diagnostic errors, procedural errors, and harm due to health care. You’ll see what I’ve said is true. Ask your favorite LLM.

AI is not a medical expert. AI is a pseudo expert. It possesses incredible a capacity to scour all of the available information and put together a coherent answer or summary. This answer or summary will greatly help the general public and physicians who are not experts in a given disease by making search, retrieval, synthesis of available information. But it’s not an expert. AI cannot be expected to know data that’s known to experts but is not yet published. It cannot know when data in published form differs from that experienced in real world practice by clinicians who see a large volume of patients with the same disease. Where the published data are wrong or exaggerate benefits or minimize risks. It cannot judge the right treatment option among similar competing treatment options (except superficially), especially based on what the patient evaluation reveals on history and examination. AI appears to be an expert in everything in the world by knowing what experts have written and made public but lacks wisdom by the very nature of how it works to produce the answer. It’s not thinking. It knows as the famous saying where the puck is but not where it’s going to be. That’s why the even the most ardent proponents of AI including the uber rich who own the models will always seek out the best human expert available for serious diseases. They may use AI to provide a summary of their disease for the expert but they are not going to mistake or substitute AI for the expert. I don’t see this changing. Because medicine is more than knowing everything that’s published or being able to retrieve it quickly. We live in a world of medicine where it’s easy to confuse pseudo experts who have gained or granted prominence with real depth of expertise and wisdom. So it’s easy to see how a lot of us are mesmerized by the speed and eloquence of AI to answer queries. Yes it does that well (and is probably sufficient 90% of the time). But as you learn how LLMs and other AI tools work you know it’s no expert, but a useful side kick. I do think it can help both experts and non experts but we must know what it’s capable of and what it’s not.

@CuriosityonX I for one welcome out 8-tentacled overlords!

🚨: Oxford biologist says that if humans go extinct, octopuses could build the next civilization.

@nic_carter AGI is probably just not a useful construct. The best models are definitely smarter than me in almost every conceivable way, and have broader knowledge than any human could ever have. But they have weird weaknesses and capability gaps where they’re super dumb.

The “it’s not AGI because machine intelligence is jagged” is dumb cope. It’s obviously AGI. If you had a friend who had a 130 IQ, could write production code flawlessly, could write academic papers of a high research caliber, pass any exam in any field with flying colors, create a sophisticate LBO model, draw technical diagrams perfectly, compose poetry in any language, and could find solutions to significant unsolved mathematical problems, you would call that person a world historical genius. Certainly, no single human has ever had intelligence that “general” before. Now you think it’s “not AGI” because it sometimes slips up and makes mistakes - so does any human that you would consider “extraordinarily intelligent.” The professor might forget a colleagues name that he has known for a decade. He is still considered intelligent. The math genius might be a little autistic and shy, unable to maintain polite conversation. Still intelligent. You might stare at the fridge for 30 seconds unable to find the butter, despite 5 million years of evolution perfecting your visual intelligence. We give intelligent humans a pass when they have jagged intelligence. So why the double standard? The qualities people list as “necessary for AGI” are important traits to have, but no longer pertain to intelligence. People will say things like “true AGI requires agency, long term goal setting, embodiment, self-direct action”. But none of those things are intelligence. Those are “things that humans have that AI lacks”. Raw intelligence, AI has it in spades. That other stuff - important yet, but broader than and different from intelligence. The unwillingness of people to acknowledge that AGI obviously exists and has existed for a while is due to a kind of anthropic chauvinism - a psychological need to believe that humans are superior in every respect, that we possess soft skills that no machine could replicate. Yes humans are different from machines, but if we are limiting the discussion solely to general intelligence, AI has it already. That battle is over. If you want to reframe the discussion to matters of human dignity and personhood, fine, but that’s not an AGI question. That’s something else. Just take the loss on AGI already. It’s over.

@Sam_kuyp She looks appropriately pompous. It’s clear it comes naturally to her. She don’t have to do no fancy book learnin’.

Plausible. Hence, more and more, the future belongs to people who find learning enjoyable and seek it out, not necessarily to people with a degree.

@leecronin Robots don’t pipette, they just produce droplets. Humans uniquely pipette.

AI does not write, it produces text. Humans uniquely write.

Fun fact: gabapentin was false and deceptively advertised when it came out, and the company paid a large fine, much of which was distributed by attorneys general to researchers (like me!) who created curricula on conservative prescribing. Gabapentin has always been a drug whose benefits were exaggerated and whose risks were minimized, and whose indication was grossly expanded beyond what was labeled. Now the butcher’s bill is being paid.

Highly recommend Dr. Grunch's videos if you are a psychiatrist or neurologist or interested in these conditions.