Gary Marcus has been tweeting about various topics related to AI recently. In one of his tweets, he mentions the brittleness of deep learning and how promptings can also be brittle. He also shares a new essay where he reevaluates the OpenAI drama in light of recent developments. Additionally, he questions the reliability of open-sourced AI and discusses Meta's beliefs and PR.
Furthermore, Gary Marcus expresses concerns about potential catastrophic consequences if open-sourced AI leads to such outcomes. He shares an original source and discusses language models' prediction abilities versus directly modeling a grounded world based on data. He criticizes someone named Cedric for their change in position and lack of response to queries, questioning their motives.
In another tweet, Gary Marcus asks if anyone has experimented with ChatGPT's progressive generation capabilities, mentioning weird anomalies like turning a muscled-up human into a croissant. He also voices his support for regulating foundation models in negotiations regarding the #AIAct.
Lastly, Gary Marcus highlights the significance of decisions made by the EU on generative AI and expresses concerns about big tech companies' lobbying efforts overriding years of hard work and the will of the people.
Overall, Gary Marcus seems to have mixed sentiments regarding AI. While he raises multiple critical points and questions certain aspects or actions within the field, it is not explicitly clear whether he holds a positive or negative view overall.
Trends mentioned in Gary Marcus's tweets include brittleness in deep learning and promptings, reevaluation of OpenAI drama, reliability concerns with open-sourced AI, discussion on Meta's beliefs and PR practices, potential catastrophe from open-sourced AI usage, anomalies in ChatGPT's progressive generation capabilities, regulation of foundation models in negotiations concerning #AIAct by EU.
Books By Professor Gary Marcus
Kluge: The Haphazard Evolution of the Human Mind
The Algebraic Mind: Integrating Connectionism and Cognitive Science