Interesting articles, August 2024

Ukraine staged a successful surprise attack into Russia itself, capturing a significant amount of territory southwest of the city of Kursk.
https://www.yahoo.com/news/thousands-flee-russia-battles-major-123112611.html
https://apnews.com/article/russia-ukraine-war-kursk-fighting-80671ef80c36b94dc1114506770cdd56

The Russians are close to seizing a strategically important city in eastern Ukraine called “Pokrovsk.” If it falls, a large section of the Ukrainian frontline will become unsupportable and will crumble.
https://www.bbc.com/news/articles/c785z8917leo

While Russia is derided in the Western press for sending poorly-trained men to fight in Ukraine, the fact is Ukraine is doing the same.
https://apnews.com/article/russia-ukraine-war-new-recruits-pokrovsk-ed2d06ad529e3b7e47ecd32f79911b83

North Korea condemned Ukraine’s invasion as a terrorist act.
https://www.reuters.com/world/europe/north-korea-condemns-ukraines-incursion-into-russia-act-terror-2024-08-18/

Russia destroyed a Challenger II tank the British donated to Ukraine.
https://youtu.be/GnYcTWuhSEA?si=gqC3UVY-XBWn8Jf6

The first F-16 fighters have been delivered to Ukraine.
https://www.twz.com/air/f-16s-arrive-in-ukraine-report

And the first Ukrainian F-16 was also destroyed during a mission to shoot down Russian cruise missiles and drones.
https://www.yahoo.com/news/ukraine-f-16-destroyed-during-200508160.html

Since October 7, 2% of Gaza’s population has been killed. That’s comparable to what some countries suffered in WWII.
https://apnews.com/article/gaza-war-hamas-dead-graves-40000-988d16b648e06e222f04964dc9440da0

Top U.S. officials say that though Israel has massively damaged Hamas, it’s unlikely that further attacks will destroy the organization. They’ve proven too adept at survival. Hamas is also so skillful at hiding the remaining Israeli hostages that rescuing them alive with commando raids is impossible. The huge civilian death toll has undermined Israel’s global image and diplomatic standing, and Hamas is so intermixed with the general population in Gaza that even the most surgical Israeli attacks will cause collateral damage.

At this point, Israel should declare victory, end military operations in Gaza, and switch to diplomacy to get its hostages back and to diminish Hamas’ remaining power as much as possible.
https://www.yahoo.com/news/gaza-israel-military-reached-end-113642816.html

‘Planted Bomb Was Used To Kill Hamas Leader In Tehran’
https://www.twz.com/news-features/planted-bomb-was-used-to-kill-hamas-leader-in-tehran-report

The U.S. Navy announced it was sending a cruise missile sub to the Persian Gulf to deter Iran from retaliating against Israel.
https://www.businessinsider.com/us-sending-powerful-submarine-middle-east-loud-warning-to-iran-2024-8

‘Mali’s northern Tuareg rebels said they had killed at least 84 Russian Wagner mercenaries and 47 Malian soldiers during days of fierce fighting in late July.’
https://www.yahoo.com/news/russians-pay-homage-wagner-fighters-125657362.html

This gives me an idea for a future weapon: robotic insects that are programmed to crawl into the fuel systems of enemy aircraft and, once they sense the aircraft have taken flight, to clog the fuel lines with some injected gluey substance.
https://www.yahoo.com/news/ntsb-investigation-said-wasps-nest-132603368.html

An Air Force investigation concluded that the crash of a $450 million B-1B bomber earlier this year was due to crew error, and that poor discipline and lax standards in their unit set the stage for the disaster. Keep in mind that machines will never forget anything, suffer from skill degradation, or need to spend any time retraining.
https://www.military.com/daily-news/2024/07/25/report-unhealthy-culture-ellsworth-air-force-base-units-contributed-b-1b-lancer-crash.html

The Indian Army’s INSAS assault rifle is, along with the British SA80, one of the worst in its class. Some of its problems owe to manufacturing defects rather than design faults.
https://www.thefirearmblog.com/blog/indian-army-plans-to-upgrade-the-insas-rifle-part-1-history-44815417
https://www.thefirearmblog.com/blog/indian-army-plans-to-upgrade-insas-rifle-part-2-my-experience-44815429

China is abandoning its “minimum deterrence” nuclear strategy.
https://www.nytimes.com/2024/08/20/us/politics/biden-nuclear-china-russia.html

The question isn’t whether the Germans COULD have won the Battle of Stalingrad (the answer is “no”), it’s whether they SHOULD have started the battle at all. They should have lurked west of the city in favorable open ground to destroy any Soviet forces that came out to fight them.

In reality, the symbolism of controlling the city bearing Josef Stalin’s name overrode military logic in Hitler’s mind, so he sent his army into a disaster.
https://youtu.be/zSah-7yvaE8?si=KEycNUPVyPM7er2r

Jacob Schiff was one of the richest men in America around the turn of the century, and a powerful advocate for Jews like himself. In the late 1800s, antisemitic violence within the Russian Empire started driving large numbers of Jews to leave. In 1903, Christian fanatics in Moldova murdered 49 Jews and raped many of their women. Schiff was outraged, and in response, he loaned Japan the equivalent of $5.3 billion in 2023 dollars–half of the country’s entire wartime military budget–to finance its war against Russia in 1905. Russia suffered a humiliating loss that resounds to this day, though Japan was teetering on the edge of bankruptcy by the end. Without Schiff, Russia might have won.
https://en.wikipedia.org/wiki/Jacob_Schiff

Pavel Durov, the founder and CEO of the social media messaging app Telegram, was arrested in France because some of the app’s users used it to commit crimes. This has major free speech implications.
https://www.bbc.com/news/articles/cx2x5yw8z7yo

‘Zuckerberg regrets bowing to Biden ‘pressure’ over Covid’
https://www.bbc.com/news/articles/czxlpjlgdzjo

A new mod for the video game Cyberpunk 2077 has been released which upgrades it graphics into being nearly lifelike.
https://www.techeblog.com/cyberpunk-2077-dreampunk-2-graphics-mod/

Google Deepmind CEO Demis Hassabis just gave an interview. Key points:

  • He freely uses the term “AGI,” which he used to be reluctant to utter.
  • He thinks AGI could be invented as early as 2030.
  • He also believes computers could invent cures for most or all diseases within 10 years, including individualized medicines.
  • In spite of that, he thinks the media is overhyping the short-term potential of the AI industry. Massive progress will happen, but slower than the media and self-interested startups imply.
    https://youtu.be/pZybROKrj2Q?si=4gHyTisaMR4SmKAb

Peter Thiel also gave an interview, and echoed Hassabis’ view that the AI industry is in a bubble. Specifically, he used the analogy that AI today is where the internet was in 1999: poised for a world-changing breakout but also in a huge bubble that will pop.
https://youtu.be/klRb0_BAX9g?si=GywdTwE3DVyw63VW

This analysis makes reasonable assumptions about growth in training data sets, data centers, electricity availability, and training efficiency, and concludes there’s no roadblock to building GPT-6 by 2030. However, a company the size of Microsoft would have to be willing to pay up to $100 billion to create it, and it would consume up to five gigawatts of electricity. (A gigawatt of electricity can power 300,000 – 750,000 homes, and an average nuclear reactor produces 1 gigawatt of electricity. [Note that one nuclear POWER PLANT can have multiple nuclear REACTORS in it.]) They could actually afford that and could even build their own power plants by the deadline.

As I wrote a few months ago, the near future of the AI industry hinges on how good GPT-5 is. If it’s a very impressive and instantly profitable product, then many big tech companies will find it worth the gamble to take the next step and build GPT-6 equivalents. If GPT-5 disappoints, then they won’t. I think GPT-5 will be released shortly after the U.S. election.

I don’t know if GPT-6 would be a “general intelligence,” but at a minimum, it would be able to replace large numbers of human jobs and to handle very complex tasks. People and organizations will get access to it through a subscription model, for example, you’ll pay $100 a month to have GPT-6 perform a job at your company that you’d have to pay a human $5,000 a month to do.
https://epochai.org/blog/can-ai-scaling-continue-through-2030

‘Our future AI overlord has determined that putting cold air inside a duct will raise the temperature of the outside of the duct above the ambient temperature of the attic.’
https://philip.greenspun.com/blog/2024/08/25/chatgpt-4o-tackles-the-challenge-of-ac-ducts-sweating-in-an-attic/

This ancient Egyptian wooden lock is so simple that only a genius could invent it.
https://youtu.be/3I25Te0qNEM?si=UN6AI-sQU4DGJDiP

A Star Trek: Deep Space Nine episode that aired in 1995 has caught up with reality.
https://youtu.be/F4x_8miSN9k?si=tCVe50ArSrL5eXdQ
https://youtu.be/Ni8LvECFoiM?si=xC1lVPDNbt5X8BVf

Stanford University mathematics professor Keith Devlin has said, “like a Shakespearean sonnet that captures the very essence of love, or a painting that brings out the beauty of the human form that is far more than just skin deep, Euler’s equation reaches down into the very depths of existence”.
https://en.wikipedia.org/wiki/Euler%27s_identity

Eight years on, and China’s project to build an international power grid has gone nowhere.
https://www.nbcnews.com/business/energy/china-unveils-proposal-50-trillion-global-electricity-network-n548376

‘According to our baseline estimates (Table 1), over the past 38 years, Chernobyl reduced the total number of NPPs worldwide by 389, which is almost entirely driven by the slowdown of new construction in democracies. Our calculations thus suggest that, globally, more than 318 million expected life years have been lost in democratic countries due to the decline in NPP growth in these countries after Chernobyl.’
https://marginalrevolution.com/marginalrevolution/2024/08/the-unseen-fallout-chernobyls-deadly-air-pollution-legacy.html

Geologists have found a place where it is possible to drill down into the Earth’s mantle.
https://www.yahoo.com/news/scientists-drilled-deep-center-earth-163300616.html

A large meteorite impact crater under Greenland’s ice sheet has been dated to 58 million years ago, squashing earlier claims that it was created 13,000 years ago and triggered the last Ice Age.
https://news.ku.dk/all_news/2022/03/giant-impact-crater-in-greenland-occurred-a-few-million-years-after-dinosaurs-went-extinct/

Published 20 years ago: ‘If intelligent life exists elsewhere in our galaxy, advances in computer processing power and radio telescope technology will ensure we detect their transmissions within two decades. That is the bold prediction from a leading light at the Search for Extraterrestrial Intelligence Institute in Mountain View, California.’
https://www.newscientist.com/article/dn6189-et-first-contact-within-20-years/

The Pentagon’s former lead UFO hunter Luis Elizondo just published an autobiography that includes a lot of detail about his old job and what the government secretly knows about aliens.
‘He claims that his D.C.-area home was “invaded,” The Times reported, by green, glowing, basketball-sized orbs. They could pass through walls and appeared to be “under intelligent control…his wife, their two daughters and their neighbors witnessed the green orbs, which they called “our friends from out of town.”’
https://www.thedailybeast.com/pentagon-alien-hunter-luis-elizondo-glowing-green-ufos-invaded-my-home

Elizondo also sat down for a long interview about himself and his book. He’s surprisingly smart.
https://youtu.be/9gLPtRwXgCM?si=nK2GJ-OZ1nQbCpOf

Crazy ways to get into space: “Lofstrom loop, StarTram and Space Cannons”
https://youtu.be/gIYpDSs8vsM?si=Dc3Jfmkoc4_f0WVo

Mars probably has liquid water several miles underground, flowing through cracks in rocks. Microbial life could be there.
https://www.cnn.com/2024/08/12/science/mars-crust-water-reservoir-insight/index.html

It’s very rare to find a dog with the right attributes to serve with special forces units or elite bodyguards. For that reason, it would probably be cheaper to keep cloning the best dogs instead of relying on the genetic crapshoot.
https://gwern.net/clone

Ten years on, and there’s been very little follow-up about this suspended animation procedure. It’s actually the norm for a supposedly revolutionary breakthrough in medicine or some other technology to rock the news media for a few days only to be never heard from again.
https://www.economist.com/technology-quarterly/2014/09/04/the-big-sleep

‘US government report says fluoride at twice the recommended limit is linked to lower IQ in kids’
https://apnews.com/article/fluoride-water-brain-neurology-iq-0a671d2de3b386947e2bd5a661f437a5

You can change the color of your eyes through plastic surgery. The only problem is it might make you slowly go blind.
https://www.cnn.com/2024/08/22/health/permanent-eye-color-change-surgery/index.html

There was controversy at the Paris Summer Olympics when gold medals were awarded to to two female boxers with genetic abnormalities that gave them male physical characteristics and male XY chromosomes.
https://www.bbc.com/news/articles/crlr8gp813ko

The FDA rejected an ecstasy-based drug that was touted as a treatment for PTSD because there wasn’t enough proof it worked.
https://www.wsj.com/health/pharma/fda-rejects-ecstasy-based-drug-cfc4b4b5

Musings 7

I think something I call a “Fake AGI” will be created in the next ten years. By simply improving existing LLMs with more data, marginally better algorithms, and bigger data centers, and then “wrapping” several specialized LLMs and computer programs together into a multi-module unit, it will become possible to build a sort of “Frankenstein” machine that would possess general intelligence. Well…most of the time.

The Fake AGI will still spit out nonsensical responses and suffer from hallucinations on occasion, sharply reminding us humans that its “mind” is fundamentally different from ours and that its intelligence is brittle. Further upgrades by its owners will roll the problem back, but never eliminate it entirely because the machine will be fundamentally incapable of general intelligence. For example, its Turing Test results will gradually improve, with it passing 99% of the time except for the 1% when it makes a totally nonsensical response that no human would. In time, its results would improve to 99.9%, then 99.99% and so on…but they would never be perfect.

But no matter how smart the machine got, no matter how well it mimicked human speech and emotion, there would still be the occasional mistakes. The strange answers and other random behaviors would be forever cited by critics as proof the machine was not really an intelligent being. Even people rejecting that stance would still admit that there was something alien about how the machine’s mind worked that we could never understand.

A “Real AGI” will require a totally different mental architecture, and several breakthrough algorithms, and will have a vastly simpler and more elegant code. I believe it is still at least 25 years away. However, from the human end user perspective, nothing might seem to change on the day the Frankenstein Fake AGI that answers correctly 99.999% of the time is switched off and the first Real AGI is switched on. The entity that they communicate with for work or pleasure will still sound the same, and the mistakes will have already become so rare that most people will have wrongly assumed the machine had been “generally intelligent” for years up to that point.  

Every human being, and probably every life form with a brain, is inherently valuable. This is because our brain structures and past experiences uniquely shape the way we process data. One person’s subjective experience and perception of something is idiosyncratic to them. When they die, that bit of individuality is forever lost. Even the life of someone as lowly as a serial killer is valuable.

Brain scanning devices like BCIs will give us unparalleled insights into how the brains and minds of humans work. In the future, once the devices are cheap and common, they could be paired with personal assistant AIs to graph the exact mental strengths and weaknesses of each individual, allowing the machines to help them maximize their potential and to learn most effectively. The brain data could also be used, along with test data, observational data, and genetic information, to make highly accurate digital clones of people. The clones could persist even after their “originals” die.