Owain_Evans(Owain Evans)
The Rationalists of the 1950s (and before) also called themselves “Rationalists”
Lives of the Cambridge polymath geniuses
Paper: On measuring situational awareness in LLMs
Paper: Teaching GPT3 to express uncertainty in words
Truthful AI: Developing and governing AI that does not lie
How truthful is GPT-3? A benchmark for language models
AI Safety Research Project Ideas
Yes, I’d also like to understand better the attraction of communism. Some off-the-cuff ideas:
It was harder to get good information about the Russian or Chinese communists during certain periods. (No Internet, fewer reliable journalists, less travel in each direction).
Non-communist countries were much more violent than post-WW2. There was more homicide and more violence that involved the state (e.g. violence in prisons, colonial violence, civil wars, interstate wars). Maybe the Soviet Union up to 1935 didn’t look radically different from non-Communist countries.
Economics was less developed and (possibly) fewer smart people had a basic grasp of the field. (Some good arguments against Marxist economics already existed but weren’t widely known).
The empirical evidence against command economies vs market-based systems was much less clear. (Modern concept of GDP was only developed in 1934!)
Non-communist countries were very conservative in some ways, e.g. rights for women and ethnic minorities, the special privileges given to the state religion, workers’ rights, availability of public education and training. Both the early Soviet Union and Communist China had some policies that were progressive relative to the status quo and to other non-communist countries.
Communists shouted about and valorized progress, modernization, industrialization, and science. (And the Soviet Union did industrialize pretty quickly and produce fairly impressive science and technology.)
Incidentally: Russell did visit the Soviet Union and came away with a negative impression. Keynes also visited and had a very negative impression. He was also in a position to evaluate Marx’s economics. Here’s Keynes’ view of Leninism taken from Wikipedia:
How can I accept a doctrine, which sets up as its bible, above and beyond criticism, an obsolete textbook [Marx’s Kapital] which I know not only to be scientifically erroneous but without interest or application to the modern world? How can I adopt a creed which, preferring the mud to the fish, exalts the boorish proletariat above the bourgeoisie and the intelligentsia, who with all their faults, are the quality of life and surely carry the seeds of all human achievement? Even if we need a religion, how can we find it in the turbid rubbish of the red bookshop? It is hard for an educated, decent, intelligent son of Western Europe to find his ideals here, unless he has first suffered some strange and horrid process of conversion which has changed all his values.
Machine Learning Projects on IDA
Paper: Tell, Don’t Show- Declarative facts influence how LLMs generalize
How do new models from OpenAI, DeepMind and Anthropic perform on TruthfulQA?
Paper: Forecasting world events with neural nets
Quantifying Household Transmission of COVID-19
Taxes in Oxford are more-or-less the same as anywhere else in the UK. These are lower than many European countries but higher than the US (especially states with no income tax).
Rent in SF is more than 2x Oxford (seems roughly right to me) but I agree with what you say on housing.
Having lived in SF and Oxford, the claim about crime and homelessness doesn’t match my experience at all (nor any anecdotes I’ve heard). I’d be very surprised if stats showed more crime in Oxford vs the central parts of SF.
PSA. The report includes a Colab notebook that allows you to run Ajeya’s model with your own estimates for input variables. Some of the variables are “How many FLOP/s will a transformative AI run on?”, “How many datapoints will be required to train a transformative AI?”, and “How likely are various models for transformative AI (e.g. scale up deep learning, recapitulate learning in human lifetime, recapitulate evolution)?”. If you enter your estimates, the model will calculate your personal CDF for when transformative AI arrives.
Here is a screenshot from the Colab notebook. Your distribution (“Your forecast”) is shown alongside the distributions of Ajeya, Tom Davidson (Open Philanthropy) and Jacob Hilton (OpenAI). You can also read their explanations for their distributions under “Notes”. (I work at Ought and we worked on the Elicit features in this notebook.)
I’d like to see discussion of data rather than mostly a priori argument (“I have a sense” … “I suspect desire”). For aggregate data, there’s SSC survey and there are studies of “ambitious” groups (e.g. the Harvard Men study, Benbow on precious math talent). There are also anecdata of the exceptionally ambitious. E.g. Musk had first child age ~30 and has many kids, Hassabis had first child aged ~29. It seems Jaan Tallinn had kids starting in his 20s before founding Skype (Wikipedia). Bezos has 4 kids (started age 37). Gates has 3 kids (started age ~40). Turing award-winners David Patterson and Judea Pearl had kids in their 20s before their biggest contributions. Yoshua Bengio in his 30s. etc