Links for 2023-11-03
Learning From Mistakes Makes LLM Better Reasoner https://arxiv.org/abs/2310.20689
RT-Trajectory: Robotic Task Generalization via Hindsight Trajectory Sketches https://rt-trajectory.github.io/
“Phind can now beat GPT-4 at programming, and does it 5x faster.” https://phind.com/blog/phind-model-beats-gpt4-fast (Caveats: "Once again, academic benchmarks like HumanEval are very, very poor proxies of in-the-wild performance. Try the model on Voyager. See if it can play Minecraft as well as GPT-4. I bet it isn’t even remotely close." https://twitter.com/DrJimFan/status/1719865206803992924)
Amazon Gears Up To Profit Mightily From The Generative AI Boom https://www.nextplatform.com/2023/10/31/amazon-gears-up-to-profit-mightily-from-the-generative-ai-boom/
With its New M3 Chips, Apple joins the AI party. https://om.co/2023/10/30/apple-launches-m3-chips-with-ai/
Google Pixel’s face-altering photo tool sparks AI manipulation debate https://www.bbc.com/news/technology-67170014
Decision theory does not imply that we get to have nice things https://www.lesswrong.com/posts/rP66bz34crvDudzcJ/decision-theory-does-not-imply-that-we-get-to-have-nice
Born deaf, this Chinese girl can hear again. A remarkable experiment takes gene therapy to a new frontier: restoring the senses. https://www.technologyreview.com/2023/10/27/1082551/gene-treatment-deaf-children-hearing-china/ [https://archive.ph/bgopA]
NIST Team Develops Highest-Resolution Single-Photon Superconducting Camera https://www.nist.gov/news-events/news/2023/10/nist-team-develops-highest-resolution-single-photon-superconducting-camera
Is Gravity a Force? Is it an Illusion? https://profmattstrassler.com/2023/10/26/the-impossible-commentary-is-gravity-a-force-is-it-an-illusion/
Sony’s High-Tech Playstation2 Will Require Military Export License [published in 1999] https://archive.nytimes.com/www.nytimes.com/library/review/061399china-chips-review.html
Mormon fertility rates are falling below replacement. https://twitter.com/NathanpmYoung/status/1717575473306255726
“The Battle of Hurtgen Forest was the worst performance and drubbing the US Army suffered in World War II. In a new essay, revisit the battle, leadership failures, and six key lessons learned that remain relevant both on battlefields and in business today.” https://nickdeiuliis.com/news/battle-of-hurtgen-forest/
You don't have to be smarter than someone to know they're wrong:
There is a Jew and a Christian who are smarter than me (e.g., Robert Aumann and Francis Collins). I can say that at least one of them must be wrong about something.
In general, rationality is not the same as intelligence, although they are correlated.
Kurt Gödel believed in ghosts. Nobel laureate Kary Mullis claims to have chatted with a glowing raccoon he met on his way to the bathroom at midnight, then lost the next six hours to alien abduction.
They are much smarter than me, but still wrong.
Someone less intelligent can sometimes tell when someone smarter than themselves is wrong by recognizing inconsistencies. For example, if a position does not make sense in light of simpler and universally accepted higher-level principles, it is likely to be wrong.
Another way to tell when someone smarter than you is wrong is to learn whether they are bound by law or social pressure to hold a particular opinion. Can someone publicly change their mind without jeopardizing their career and social connections, such as friendships and family relationships?
But this is not always possible. Because the smarter someone is, the easier it is for them to rationalize ideas that make no sense. If you are really smart, you can invent valid arguments or cobble together concepts and ideas to defend your cherished beliefs. The result can be an intricate argumentative framework that shields you from any criticism, yet seems perfectly sane and rational from the inside.
e^(i*pi) = -1
(e^(i*pi))^2 = (-1)^2 = 1= e^(i*2*pi)
e^(i*2*pi) = e^0
ln(e^(i*2*pi)) = ln(e^0)
i*2*pi = 0
In essence, this can be illustrated by the relationship between adults and children. Adults can be confused by more complex ideas than children. Children, on the other hand, can be infected by the same ideas as adults. Similarly, a superhuman AI might be able to reason its way out of a box by convincing its gatekeeper that it is rational to do so. But that doesn't make it true.
Ukraine links:
Ukraine’s top general on the breakthrough he needs to beat Russia https://www.economist.com/europe/2023/11/01/ukraines-top-general-on-the-breakthrough-he-needs-to-beat-russia [https://archive.ph/FqTYm]
The commander-in-chief of Ukraine’s armed forces on what he needs to beat Russia https://www.economist.com/by-invitation/2023/11/01/the-commander-in-chief-of-ukraines-armed-forces-on-what-he-needs-to-beat-russia [https://archive.ph/dJDI2]
"Modern Positional Warfare and How to Win in It" by Gen. Zaluzhnyi, Commander-in-Chief of AFU https://infographics.economist.com/2023/ExternalContent/ZALUZHNYI_FULL_VERSION.pdf
Episodes of assault actions of soldiers of the 3rd Separate Assault Brigade in the Bakhmut direction https://twitter.com/small10space/status/1720002742519824885
“It's been almost exactly a year when Russians tried to storm Vuhledar for the first time and even a year later the results are the same: disastrous.” https://twitter.com/NOELreports/status/1720108242120974757
🔥 MARIINKA DIRECTION | RUSSIAN LOSSES https://twitter.com/GloOouD/status/1720207094266138766
The Battle of Avdiivka documented by Ukraine's 47th Mechanized. https://twitter.com/IAPonomarenko/status/1720194072361521447