JH Kim joins the Chem Dept in Korea University in 1996, fresh faced, 24 years old.
He’s a synthesist, an experimental chemist of the old school. He believes the truth is in the making, and that you follow the truth even when it destroys your reality.
A squiggle on a chart, a faintly heard echo of something real happening.
Kim plays Marco Polo with the universe, blindfolded, calling out again and again with each experiment, listening on the ESR . Sometimes warmer, sometimes colder. He invents entire new… https://t.co/Nqs67Cc7gttwitter.com/i/web/status/1…
⏰Previously, on Friday 28 July
Kwon goes rogue and presents at the MMC conference.
Lee is forced to give an interview to Yonhap, retracting the 3 author paper, and disclosing Kwon had been fired 4 months earlier.
YH Kwon goes rogue on Friday, showing up to the Metallic Multilayers Conference in Seoul unannounced.
He presents the data from the to be retracted 3 author paper, affiliating himself with both the Qcenter and Korea University. Both institutions declare him… https://t.co/3LExkO2u9Btwitter.com/i/web/status/1…
OpenAI admits that all academic benchmarks should be assumed to have been memorized by GPT.
Their test criteria now draw from human psychology and they build novel and difficult questions. This is where the ChatGPT usage data is going to be very useful for next gen. @MikePFrank
AI work in academia, 🇨🇳 , Google and other places not able or willing to get consumer feedback is screwed. AGI training memorizes all benchmarks eventually.
You have to have humans continually generate new and novel questions.