THANK YOU BOTH: VERY INSPIRATIONAL JUST SENT THIS TO MY SON DOING HIS PhD AT CARNEGIE MELLON IN THIS FIELD
Fascinating interchange and very erudite commentaries about AI for research. I found the following specific comments esp relevant to our novel biochemical research relative to the use of the repurposed drug hydroxyurea linking critical neurotransmission a7NAChRs to Covid and possibly for Post-acute sequelae of COVID19, namely:
“So like you mentioned, it's getting harder and harder to get timely feedback from human experts. Number of papers just growing so quickly, and especially if you're thinking about a junior researcher, maybe a students or someone from an under-resourced university or settings, it's especially hard for them to get timely feedback from experts in the field.”
Given the significant barriers for publishing novel research in major journals with inevitable biases by lower level screening editors, many important (and urgent) discoveries are being buried unnecessarily. The preprint concept as a pre-peer review mechanism could be one solution esp if it also utilizes an AI component as promoted by QEIOS. Ref: https://www.qeios.com/
AI functions as a screener for expertise and the experts critique the proposed article adding both accuracy as well as validation to
the authors proposed publication. The process appears to be a hybrid and may actually reduce or at least minimize false or faulty information utilizing AI in screening certain topics from the world’s literature. I experienced that and was informed by a credentialed medical librarian that one article I located thru an AI search tool simply did not exist. It was dis-information at its finest!
I envy him. Its a wonderful time to be alive and as a scientist, curious with all the different tools of investigation available to you. Many of the answers lie so close within the scientific grasp despite the vast unknown. People like James are just the kind of examples our young need to inspire them. As indeed you have been sir, to many who have followed yours....kudos
Thank you both for the multiple revelations in this podcast. I have thought twitter was underused as a training data source, but I had no idea you could train on pathology slides from these images and text. Also brilliant way to diagnose the deterioration of GPT-4 over time by using smaller surrogate models. I need to start following James Zou more carefully!