the best competitive programmer in the world (gennady korotkevich, aka tourist) recently crossed the 4000 ELO barrier in Codeforces. o1 is about 1807 ELO.
the best ai model is compared against the best human in the context of competition programming, to set a clear standard of comparison.
similarly, terence tao represents the highest levels of math in analysis.
his input is valuable in regards to math. his summary of the current capabilities of o1 is important because we can then understand the level of competence the best ai models have right now, and set a standard of comparison just like with coding.
site note:
any number of phds = not the same expertise.
there are thousands of phds who graduate every year, let alone thousands of unemployable phds who fail to get a professorship.
there are only 2-4 fields medalists chosen every 4 years.
there is a good reason why the committee chooses older researchers with seemingly outdated, decades-old contributions over younger researchers.
you can't award the nobel posthumously. in other words, once someone dies, they take their life's work with them to the grave without any further formal recognition. so older folks are prioritized because everyone knows that the younger folks still have long lives ahead of them, and their time to shine will come.
also, the reason why nobel laureates in economics tend to be older than the other sciences is because of the lag time of proving important contributions.
in the basic sciences, new discoveries can be overnight. of course, incremental progress takes decades as well, but the culmination --- the breaking point of a new discovery --- can be instantaneous.
in economics, which is better categorized as speculation of rational and irrational human decision-making, it may take decades to prove a theorem (especially in macro). there is no "overnight success". you can say the most absurd or correct claim such as "new neoclassical synthesis is the best monetary framework!" but even if it were true, it doesn't matter unless there is substantial evidence. and that usually takes decades because of long-term business cycles and novel, unprecedented crises.
the consequence of this lag time between when the seminal paper is published and when there is consensus in the field is massive.
example:
paul romer won the econ nobel in 2018. but the paper he wrote that earned him the nobel was actually his PhD thesis from all the way back in 1983, when he 29. he hypothesized that one of the most important contributions to long-term economic growth was ideas, and how it is free to distribute and re-use innovative ideas.
even if his hypothesis were true in 1983, it wouldn't matter unless there was substantial evidence of it. and you can be sure that the rapid growth many developing countries experienced post-1980s and onwards helped to support his claim.
at 28 the most you'll make at big tech is $500k right now.
there are people who are raking in $10M+ per year in their 20s, and they ain't thinking about charity or retiring.
the best way to improve the world is to create new businesses that innovate in niche areas or pioneer brand-new industries, thereby employing thousands of people in the process and raising GDP per capita more than you could have as an employee.
then when you get rich and "make it", you can create your own philanthropy and donate if you so choose.
airbnb pays more than FAANG and is fully remote. senior engineer above $400k and staff engineer close to $600k. i hope the cities don't crack down on my employer :(
fully remote without pay reduction + top pay + great WLB (less than 30 hours/week) = awesome job.
taking away airbnb's eligibility would just be disastrous for the employees who worked hard to get in
the best competitive programmer in the world (gennady korotkevich, aka tourist) recently crossed the 4000 ELO barrier in Codeforces. o1 is about 1807 ELO.
the best ai model is compared against the best human in the context of competition programming, to set a clear standard of comparison.
similarly, terence tao represents the highest levels of math in analysis. his input is valuable in regards to math. his summary of the current capabilities of o1 is important because we can then understand the level of competence the best ai models have right now, and set a standard of comparison just like with coding.
site note: any number of phds = not the same expertise. there are thousands of phds who graduate every year, let alone thousands of unemployable phds who fail to get a professorship.
there are only 2-4 fields medalists chosen every 4 years.