Things I read last week… (WB 31 Jan 2022)
“we created a system called AlphaCode that writes computer programs at a competitive level. AlphaCode achieved an estimated rank within the top 54% of participants in programming competitions by solving new problems that require a combination of critical thinking, logic, algorithms, coding, and natural language understanding.”
DeepMind’s AlphaCode can code at an average human level.
OpenAI builds a neural theorem prover that solves some high school Math Olympiad problems.
“There are approximately seven scanners per million inhabitants and over 90% are concentrated in high-income countries. We describe an ultra-low-field brain MRI scanner that operates using a standard AC power outlet and is low cost to build.”
A low cost MRI scanner.
How a PhD astrophysicist thinks about data.
How mmWave radar enables ambient computing.
'Quantum friction' slows water flow through carbon nanotubes.
“… it’s vital to keep research human. Reality is multidimensional, and people are skilled at looking at issues from multiple angles, following their curiosity even when it doesn’t seem at first to have a direct, linear connection to the bigger situation. ”
The McNamara Fallacy: Measurement is not understanding
Bank of England blog on the global spillovers of the Fed information effect.
Jeff Dean, the head of Google's AI efforts, explains neural networks. [YouTube]
3Blue1Brown on ‘The mathematically optimal Wordle strategy’. [YouTube]