Links for 2023-05-09
We need a Manhattan Project for AI safety and alignment https://www.politico.com/news/magazine/2023/05/08/manhattan-project-for-ai-safety-00095779
This AI Hunts for Hidden Hoards of Battery Metals https://spectrum.ieee.org/ai-mining
“Humans excel in solving complex tasks through a mental process of moving from one idea to a related one, and the progress is made through milestones or subgoals. Taking advantage of this property, we propose Adaptive Subgoal Search (AdaSubS), a search method that adaptively adjusts the planning horizon. To this end, AdaSubS generates diverse sets of subgoals at different distances. A verification mechanism is employed to filter out unreachable subgoals swiftly, allowing to focus on feasible further subgoals.” https://sites.google.com/view/adaptivesubgoalsearch/home
Shap-E: Generating complex and diverse 3D assets in a matter of seconds https://arxiv.org/abs/2305.02463
Plan, Eliminate, and Track -- Language Models are Good Teachers for Embodied Agents https://arxiv.org/abs/2305.02412
Lift Yourself Up: Retrieval-augmented Text Generation with Self Memory https://arxiv.org/abs/2305.02437
ZipIt! Merging Models from Different Tasks without Training https://arxiv.org/abs/2305.03053
Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision https://arxiv.org/abs/2305.03047
Terence Tao: There’s more to mathematics than rigour and proofs https://terrytao.wordpress.com/career-advice/theres-more-to-mathematics-than-rigour-and-proofs/
The Existential Risk of Math Errors: "Any useful discussion will be rigorous, hopefully with physics and math proofs; but proofs themselves are empirically unreliable...this sets a basic upper bound on how much confidence we can put on any belief." https://gwern.net/math-error (see also: https://axisofordinary.substack.com/p/links-for-2021-12-26)
Confidence levels inside and outside an argument: "One must distinguish between levels of confidence internal and external to a specific model or argument." https://lesswrong.com/posts/GrtbTAPfkJa4D6jjH/confidence-levels-inside-and-outside-an-argument
There are an infinite number of proofs that there are an infinite number of primes https://blog.computationalcomplexity.org/2023/05/there-are-infinite-number-of-proofs.html
Why the U.S. should fight Cold War 2 https://www.noahpinion.blog/p/why-the-us-should-fight-cold-war
What does superintelligence mean? Here are four properties: https://twitter.com/CRSegerie/status/1655560534563909633
See also: Why human intelligence is just a glimmer of what is possible https://axisofordinary.substack.com/p/why-human-intelligence-is-just-a
Sex-Biased Mental Disorders: https://link.springer.com/article/10.1007/s00239-022-10072-2