Announcements
Y’all: I got my new passport! I had to spend 3 days on the bus to go to Miami and back, but at least it taught me that I now no longer choke myself if I sleep without my CPAP machine!
Weekly Shoutout
Shoutout again to @Keanu for the epic assist in helping me see @ThePianist this weekend in LA!
Also huge shoutout to @William for letting me crash with him in Venice!
Food for Thought
Regardless of your definition of Superintelligence, doesn't the fact that AI is now better than most undergraduate students a sign that we are close to building AGI? Let's review the key benchmarks showing the next phase transition in evolutionary history!
ChatGPT has literally passed the Bar exam now!
GPT is also on its way to becoming a doctor!
Beyond language tasks however, deep learning systems have also reached human performance across a variety of image recognition tasks
Despite the $100m price tag for training Large Language Models, as a broader trend training AI systems has gotten cheaper and cheaper over the last decade
Lastly, looking at the ELO rating of Chess AIs shows us just how exponentially more powerful these models are becoming compared to even peak human performance
Crowd Pleasers
Book bans are back, but this time it's not from the Nazis but from Republicans performing their own form of cancel culture, with Texas leading the way
What are these conservatives banning? Mostly novels about gay teens...bc books be grooming them kids ahahha
Mind Benders
The Kardashian Index (K-Index), named after Kim Kardashian, is a satirical measure of the discrepancy between a scientist's social media profile and publication record. Proposed by Neil Hall in 2014, the measure compares the number of followers a research scientist has on Twitter to the number of citations they have for their peer-reviewed work.
Jokes aside, this index suggests an interesting meta-science question: what happens if we analyze scientists buy plotting the # of peer-reviewed papers against their popularity as measured by Google search results?
# of peer reviewed papers is interesting. But what about # of citations from other papers? Didn’t you say that was a good metric ?