Month: February 2026

Posted on

Interesting articles, February 2026

The U.S. and Israel have assembled a massive air and naval strike force in the Middle East and have attacked Iran. The fighting is ongoing as of this writing.
https://www.cnn.com/world/live-news/israel-iran-attack-02-28-26-hnk-intl
https://www.twz.com/news-features/final-pieces-moving-into-place-for-potential-attack-on-iran
https://www.wsj.com/world/middle-east/in-a-first-u-s-deploys-combat-jets-to-israel-for-potential-wartime-mission-in-iran-c739d870

Here’s incredible footage of an intercepted Iranian missile crashing to the ground in Qatar.
https://youtu.be/VDk5cPwTPKY?si=M5yNX_1UMz4BsMeW

‘Iran strikes threaten to deplete US weapons supplies — and put American troops at risk’
https://www.politico.com/news/2026/02/25/iran-weapons-trump-troops-defense-00797801

Colombia has a large population of recently discharged soldiers who have become mercenaries in Ukraine and other conflict zones because they can’t find jobs at home.
https://www.economist.com/the-americas/2026/02/04/why-so-many-colombians-fight-in-foreign-wars

‘Russia killed opposition leader Alexei Navalny using dart frog toxin, UK says’
https://www.bbc.com/news/articles/cwyk4lz4e3eo

A Norwegian scientist built one of the hypothesized microwave weapons responsible for “Havana syndrome,” fired it at his head, and now has the syndrome. The U.S. government also secretly bought a similar microwave weapon on the black market from an undisclosed country and found Russian components inside of it.
https://www.washingtonpost.com/national-security/2026/02/14/havana-syndrome-cia-norway-experiment/

The expiration of the New START treaty between Russia and America will free the U.S. to quickly beef up its nuclear arsenal.
https://www.twz.com/air/usaf-ready-to-make-all-b-52s-nuclear-capable-load-icbms-with-multiple-warheads-if-directed

One reason the U.S. has so many nuclear weapons is it must be prepared for a war where Russia and China are both against us, and our allies offer little help. Now that China is expanding its nuclear arsenal, the U.S. will need to do the same. The 40-year trend of nuclear disarmament is over.
https://www.economist.com/international/2026/02/03/america-risks-a-nuclear-arms-race-with-china

Russia is using an intermediate-range ballistic missile against Ukraine that it illegally developed 10 years ago. During his first term, Trump withdrew the U.S. from the Intermediate-Range Nuclear Forces (INF) Treaty because U.S. intelligence had discovered Russia had the treaty-banned missile. It can carry conventional or nuclear warheads.
https://www.reuters.com/business/aerospace-defense/ukraine-images-indicate-russia-used-missile-heart-nuclear-pact-collapse-2026-02-26/

Early Soviet warships were low-tech and full of design flaws.
https://youtu.be/1CWLrpYGKVE?si=v_MxRhIH1iF8J_kc

A Chinese video generator has made waves after it was used to make a real-looking CGI clip of Tom Cruise and Brad Pitt fighting in a movie. It’s yet another milestone on the road to personalized entertainment content.
https://www.hollywoodreporter.com/business/business-news/mpa-cease-and-desist-bytedance-seedance-2-0-1236510957/

Here’s an incredible Seedance 2 CGI video of Neo and Agent Smith fighting.
https://youtu.be/c00ZNn9v01o?si=UIBMePPDAzSeuplM

Elon Musk is in legal trouble in France for allowing Grok to undress images of people (mostly women).
https://apnews.com/article/france-x-investigation-seach-elon-musk-1116be84d84201011219086ecfd4e0bc

‘It is impossible to draw meaningful conclusions from METR’s Long Tasks benchmark — in particular once one realizes that its numerous flaws are probably compounding in unpredictable ways.’
https://www.transformernews.ai/p/against-the-metr-graph-coding-capabilities-software-jobs-task-ai

A few X posts about AI:

McKinsey agrees with me that AI will make capitalism more efficient and cutthroat than ever.
https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-automation-curve-in-agentic-commerce

The Atlas robot can now do somersaults. I like that they showed some of the outtakes.
https://youtu.be/UNorxwlZlFk?si=MFEt7uM72a4iU1Ik

“An LLM-controlled robot dog saw us press its shutdown button, and the LLM rewrote the robot’s code so it could stay on.”
https://palisaderesearch.org/blog/shutdown-resistance-on-robots

The 2026 Spring Festival Gala by China Media Group featured an incredible, choreographed robot dance performance.
https://youtu.be/mUmlv814aJo?si=37YQ1QT8YhE5_B_g

Rodney Brooks issues his yearly technology predictions. He says LLMs will not lead to AGI, and that the LLM era of what I call “Fake AI” will last at least 10 more years. Also, humanoid robots have also been oversold the the media and are still too dangerous and limited for use in the real world.
https://rodneybrooks.com/predictions-scorecard-2026-january-01/

‘Amazon shelves Blue Jay warehouse robot’
https://www.foxnews.com/tech/amazon-shelves-blue-jay-warehouse-robot

Twitter founder Jack Dorsey just laid off 40% of his staff at his new tech company, “Block,” because AI made them redundant.
https://apnews.com/article/block-dorsey-layoffs-ai-jobs-18e00a0b278977b0a87893f55e3db7bb

The “AI 2027” report was published last April and predicted AGI would be created by 2028. Now that a whole nine months have passed, the authors have been able to compare actual tech developments with what they predicted, and unsurprisingly, things are moving slower than they hoped, so they have bumped their date for the rise of AGI to “mid-2028 to mid-2030.” I think it’s wise to accept the upper bound.

I wonder if, after another nine months of slower-than-hoped tech progress, they will push back their big prediction again. Ultimately, and in spite of its apparently rigorous methodology, the “AI 2027” report might turn out to have been useless.
https://blog.ai-futures.org/p/grading-ai-2027s-2025-predictions

A wonderful essay that contains ideas I’ve written about on this blog: “Post-AGI Economics As If Nothing Ever Happens”
https://www.lesswrong.com/posts/fL7g3fuMQLssbHd6Y/post-agi-economics-as-if-nothing-ever-happens

This essay, “The 2028 Global Intelligence Crisis”, predicted dire economic impacts from AI by 2028, and actually caused a measurable drop in global stock markets. It’s funny because all of the essay’s core ideas are predictions I made years ago, and shared here. If only more people listened to me!
https://www.citriniresearch.com/p/2028gic

Here’s are two counterpoints to the essay:
https://marginalrevolution.com/marginalrevolution/2026/02/is-there-an-aggregate-demand-problem-in-an-agi-world.html
https://www.economist.com/finance-and-economics/2026/02/25/a-viral-research-note-on-ai-gets-its-economics-wrong

Sam Altman made a controversial yet entirely factual statement that downplayed the energy requirements of data centers by pointing out the vast energy required to support one human into adulthood. This is a very good point. As Noam Chomsky famously observed and other scientists expanded upon, the human brain comes hardwired with many algorithms for intelligent thinking that only came into existence over tens of millennia of evolution. In fact, it probably actually the product of millions of years of evolution that began with our pre-human ancestors first diverging from other mammals.
https://www.facebook.com/reel/931074852795152

Elon Musk is making huge, risky bets on household robots, autonomous taxis, near-term sharp improvement in AI, and data centers in space. He’s having to marshal all of his already vast resources to give it a shot.
https://www.economist.com/business/2026/02/04/elon-musk-is-betting-his-business-empire-on-ai

Anthropic is refusing the Department of War’s demand to let it use its AI systems for mass surveillance and autonomous weapons. I like how the head of the company acknowledges that the U.S. will SOMEDAY need a fully autonomous army to defend itself. As in every other realm of endeavor (business, art, whatever), AGI and robots will render humans obsolete, and any organization that wants to stay competitive will need to switch to machines. His important caveat is that the machines aren’t smart enough to do it yet, and so can’t be trusted.

‘Partially autonomous weapons, like those used today in Ukraine, are vital to the defense of democracy. Even fully autonomous weapons (those that take humans out of the loop entirely and automate selecting and engaging targets) may prove critical for our national defense. But today, frontier AI systems are simply not reliable enough to power fully autonomous weapons.’
https://www.anthropic.com/news/statement-department-of-war

‘Pennsylvania State Police were stymied in their investigation into the violent rape of a woman in 2016 on a remote cul-de-sac outside Milton, a small community in the center of the state. With no clear leads, police obtained a warrant directing Google to disclose accounts that searched for the victim’s name or address over the week when she was attacked.’
https://apnews.com/article/google-reverse-keyword-search-privacy-c5a0bc6f3790213f92e78aae720d2379

‘Women were significantly more likely to focus on areas where danger could lurk, including unlit areas, potential hiding spots and places where they might be trapped, often off to the side of their path.’
https://www.deseret.com/2024/2/7/24065126/byu-study-gender-difference-night-time-walking-jogging-vigilance/

‘There is an unfortunate bias against attractive people. It stems from the belief that their looks allow them to glide through life and provides them with opportunities that average-looking people don’t have. The doctors I knew were always curious when meeting a good-looking sales rep: Is there any substance behind the pretty face?’
https://medicalsalestraining.com/the-risk-and-myth-of-the-good-looking-sales-rep-in-medical-sales/

More evidence that marijuana inflames some kinds of latent mental illnesses.
https://www.npr.org/2026/02/21/nx-s1-5719338/cannabis-marijuana-weed-teens-psychosis-jama

A small advance towards being able to freeze and thaw out human brains without destroying them has been made.
https://www.biorxiv.org/content/10.64898/2026.01.28.702375v1

Posted on

Musings 11

An AGI Dyson Swarm will have problems synchronizing its satellites thanks to light speed latency, and this will have major implications for its intelligence and consciousness. A Dyson Swarm is a vast cloud of solar-powered satellites that surround a star in a spherical formation. Each satellite converts light into electricity and then uses it for various tasks. Doubtless, a Swarm would be comprised of many types of satellites that are specialized for different functions, like data processing, communications, or manufacturing space ships. The radius of the Swarm would need to be enormous since the satellites would melt or at least run inefficiently if too close to the star. A radius equal to the Earth’s distance from the Sun, 1 astronomical unit (AU), isn’t unreasonable.

The satellites responsible for data processing will be distributed and will need to communicate with each other, just as the neurons in your brain need to send electrochemical signals to their neighbors to collectively produce intelligent thought. However, if the Swarm has a 1 AU radius, then it will take 17 minutes for a light-speed signal to travel from one end of it to the other, plus another 17 minutes to get any response. While acknowledging the existence of lag within the human brain (e.g. – scientists have measured how long it takes for signals to travel across the brain, and you instinctively understand that it takes time to think to arrive at a solution to a problem), this will be much different: Consider what the processing speeds will be on the futuristic computers emplaced in each satellite. Each computer will be “thinking” so fast that their subjective experience of time will probably be more drawn-out than it is for humans, making a 17-minute communication lag feel like an eternity.

It occurs to me that this discordance would either prevent a Dyson Swarm from having a unitary, coherent consciousness, or give rise to a consciousness that would be very alien from our own (this compounds upon whatever fundamental alienness there is to machine consciousness compared to human). In the former case, it would ensure the Dyson Swarm were comprised of many AGI individuals.

A future step in tech-enhanced human evolution will be dividing our consciousnesses into many directions at once, in turn allowing one person to control multiple bodies simultaneously. One consequence of this is that you’d be able to have sex with yourself.

Autonomous cars will help people easily earn side cash delivering packages. Right after you get into a car and enter your destination, the car’s computer will be able to determine if anyone needs a package delivered from one point to another along your intended route. It would instantly tell you if there were any opportunities to transport packages (or people) for profit and let you decide whether it was worth it. 

There will also be autonomous cars not designed to carry humans. For example, imagine something smaller than a SmartCar that lacks a steering wheel or seats and is solely designed to deliver food. It might just have one or two little doors that would pop open so a customer could reach inside to get their order.

Thanks to better reaction speeds, machines will be able to drive cars safely under conditions that many humans couldn’t handle. For example, autonomous cars could drive at high speeds while separated from each other by only a few feet. They would coordinate their movements so none of them changed velocity without giving the others time to react. This would allow vehicles to pack together more densely, effectively increasing the capacities of existing roadways.

The losers of the autonomous car revolution include:

  1. Car insurance companies. Fewer accidents means less need for insurance. The industry will be forced to consolidate and to focus more on selling other products, like homeowners insurance (though if everyone also has a robot butler that can put out fires and fix major plumbing leaks, demand for that type of insurance also drops).
  2. Traffic police
  3. Local governments that get revenue from red light cameras, speed cameras, and ticketing drivers for infractions.  
  4. Home movers. In the future, you might pay some local guys off Craigslist to load your furniture and other belongings into an autonomous truck, then it would drive itself to the destination, where you’d pay a second group of local guys to unload it. Moving a full house worth of stuff halfway across America can easily cost $10,000 today. The system I described might cut that down by 70-80%. 
  5. The car moving industry (e.g. – someone else drives your car across country for you because you moved) will also die out once cars are autonomous.
  6. Owners of large parking lots, particularly around airports.

If intelligent aliens are visiting Earth, one way they would monitor us is with machines that are the size and shape of insects. It might be that there’s an alien database of DNA from every human who has ever lived, compiled thanks to billions of robot mosquitoes sucking blood from people while they slept. Maybe they’ve got DNA from a beloved pet you had as a child. And depending on how long the aliens have been here, there might even be samples of DNA from extinct species like dinosaurs.

If, generally speaking, consumer technologies and AI will increasingly focus on satisfying individual preferences (e.g. – virtual reality game where you’re the only player and you’re always #1), then why should it extend to the foods and drinks you consume? Continuous surveillance will allow your personal assistant AI to quickly learn your food and drink preferences. Once robot chefs are everywhere, you’ll be able to have your exact favorite meal or sandwich you remember from your childhood anytime. Even if you went into a restaurant or someone else’s house, it would be a simple matter for you to transmit your recipe to the robot chef who was there and have it make something for you, perfectly.

A small step towards this future can be seen in the newer restaurant soda machines that can mix dozens of syrups to make hundreds of kinds of drinks.

With better knowledge of how the human flavor palate works (there is probably some genetic component to this), machines could go a step farther by tweaking your favorite recipes in ways that will make them taste even better, and by recommending new foods and drinks you wouldn’t have tried but will probably love.