Eric Horvitz has been tweeting about a variety of topics lately. He recently collaborated with Abi Sellen on a study about designs for human-AI interaction, drawing lessons from aviation and human factors research. He also had a good conversation with European broadcasters about the directions and influences of AI in the media. Eric has been actively promoting summer and spring internships at Microsoft Research, which cover various areas including neural network models, human-AI collaboration, health & bioscience, and the societal impacts of AI. Additionally, he shared a podcast episode he enjoyed with Molly Wood at WorkLab.
One trend that Eric is speaking about in his tweets is the power of prompting in unleashing expertise from GPT-4 on medical benchmarks without additional fine-tuning or expert-curated prompts. He highlights how GPT-4 performed well on medical challenge benchmarks with simple prompting and explores the potential benefits of more systematic exploration of prompting techniques. This trend showcases advancements in AI language models for medical applications.
In terms of sentiment analysis, it is difficult to determine an overall positive or negative sentiment based solely on these tweets. Eric seems enthusiastic about his collaborations and research findings but does not explicitly express positive or negative opinions regarding the direction of AI as a whole.
In summary, Eric Horvitz's recent tweets cover various topics such as designs for human-AI interaction, internships at Microsoft Research, discussions on AI in the media, and advancements in prompting techniques for medical benchmarks using GPT-4.
SOME AI BOOK RECOMMENDATIONS