Links for 2022-12-02
“What is the war in Ukraine teaching Western armies? This report seeks to outline key lessons, based on the operational data accumulated by the Ukrainian General Staff, from the fighting between February and July 2022.” https://rusi.org/explore-our-research/publications/special-resources/preliminary-lessons-conventional-warfighting-russias-invasion-ukraine-february-july-2022
Thread about neurotech startups https://threadreaderapp.com/thread/1597583519676792832.html
"first processor core implementing the posit standard in hardware & showed that, bit-for-bit, the accuracy of a basic computational task increased by up to four orders of magnitude, compared to computing using standard floating-point numbers" https://spectrum.ieee.org/floating-point-numbers-posits-processor
Scientists have observed a single-cell alga evolve in real-time into a multicellular organism. The transition took around a year and was caused by the introduction of a predator into the environment. https://nature.com/articles/s41598-019-39558-8 (for more see here: https://axisofordinary.substack.com/p/recent-findings-that-made-the-fermi)
The hunt for habitable planets may have just gotten far more narrow: “The pressure from a class M red dwarf star’s radiation is immense, enough to blow a planet’s atmosphere away,” boding poorly for the “Goldilocks Zone” around such stars (the most common type.) https://www.cnn.com/2022/10/25/world/exoplanets-search-narrowed-scn/index.html
A thought experiment: what if we talked about the over-60s’ screen time the way we talk about young people’s screen time? https://webdevlaw.uk/2022/08/18/a-thought-experiment-what-if-we-talked-about-the-over-60s-screen-time-the-way-we-talk-about-young-peoples-screen-time/
"humans displays significantly higher levels of fearfulness than chimpanzees, bonobos and orangutans...the fearful ape hypothesis proposes that...enhanced fearfulness is assumed to be of adaptive value due its functions in facilitating cooperative care" https://cambridge.org/core/journals/behavioral-and-brain-sciences/article/human-fear-paradox-affective-origins-of-cooperative-care/B6DC0651D8B6EC81C43AF41BC22EDD15
The case for a pandemic early-warning system https://www.city-journal.org/the-case-for-a-pandemic-early-warning-system
“When can **Equilibrium Models** learn from simple examples to handle complex ones? We identify a property — Path Independence — that enables this by letting EMs think for longer on hard examples.” https://arxiv.org/abs/2211.09961
New CRISPR-based tool inserts large DNA sequences at desired sites in cells https://news.mit.edu/2022/crispr-gene-editing-dna-1124
Collaborative Creative Writing with OpenAI’s ChatGPT https://andrewmayneblog.wordpress.com/2022/11/30/collaborative-creative-writing-with-openais-chatgpt/
For reference: https://ai.googleblog.com/2022/06/minerva-solving-quantitative-reasoning.html?m=1
"Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM."
"Evaluating our model in 2022 Poland’s National Math Exam it performed above the national average, it solved more than 80% GCSE Higher Mathematics problems and evaluated a variety of STEM undergraduate problems from MIT, it solved nearly a third of them."
Cf. In 2016, surveyed AI experts predicted that "Perform as well as the best human entrants in the Putnam competition" would take 34 years to achieve, in 2050: https://arxiv.org/pdf/1705.08807.pdf#page=16
More on ChatGPT:
ChatGPT: *writes Shakespeare sonnet explaining Bubblesort*
Skeptic: Meh! No Iambic pentameter. Wrong rhyme structure.
Prompt engineering is the new hacking:
Which will get really interesting when the AIs start designing prompts to hack humans.