The Gradient: Perspectives on AI

Jonathan Frankle, Chief Scientist at MosaicML and a researcher in sparse neural networks and technology policy, shares insights on the lottery ticket hypothesis, measurement in neural networks, scaling laws, and the acquisition of Mosaic by Databricks. He also discusses the impact of AI on society and the importance of asking fundamental questions about policy and regulation.

"The Gradient Podcast" features Nao Tokui, an artist/DJ and researcher known for using AI in music production. They discuss leveraging AI in music creation, their collaboration with Nujabes, exploring originality in generative AI systems, enhancing DJing with AI, and the concept of 'surfing with AI' in the creative process. They also highlight the potential of AI as a creative tool and the importance of embracing unexpected outcomes.

Divyansh Kaushik, Associate Director for Emerging Technologies and National Security at the Federation of American Scientists, discusses the realities of AI policy. They explore AI regulation, bringing AI talent into the US, US-China relations, China's AI regulations, and the challenges of AI policymaking and regulation.

Tal Linzen, Associate Professor of Linguistics and Data Science at New York University and a Research Scientist at Google, discusses the intersection of psycholinguistics and deep learning, the challenges of evaluating language models, the limitations of current models in representing human language capabilities, and the potential for using neural network representations to encode compositional structure.

Kevin K. Yang, a senior researcher at Microsoft Research, discusses the intersection of machine learning and biology in protein engineering. Topics include generative models for proteins, protein structure generation using folding diffusion, protein evolutionary dynamics, limitations of interpretability in AI, and future directions in protein engineering.

Arjun Ramani and Zhengdong Wang discuss the challenges of achieving transformative AI. They explore topics such as bottlenecks to progress, deployment challenges in high-stakes settings, the importance of synthetic data, and the complexities of automating tasks. They also touch on the impact of AI on economic growth and the unpredictability of implementing AI in different sectors.

In episode 90 of The Gradient Podcast, Daniel Bashir speaks to Miles Grimshaw.Miles is General Partner at Benchmark. He was previously a General Partner at Thrive Capital, where he helped the firm raise its fourth and fifth funds, and sourced deals in Lattice, Mapbox, Benchling, and Airtable, among others.Have suggestions for future podcast guests (or other feedback)? Let us know here or reach us at editor@thegradient.pubSubscribe to The Gradient Podcast:  Apple Podcasts  | Spotify | Pocket Casts | RSSFollow The Gradient on TwitterOutline:* (00:00) Intro* (02:48) Miles’ background (note: Miles is now the second newest GP at Benchmark)* (06:07) Miles’ investment philosophy and previous investments* (12:25) Investing in the “decade of deep learning” and how Miles became interested in AI* (18:53) Miles’ / Benchmark’s investment in Langchain* (24:29) On AI advances and adoption* (39:25) Hardware shortages, radically changing UX for LLMs* (48:12) Opportunities for AI applications in new domains* (50:15) Miles’ advice for potential founders in AI* (1:00:00) OutroLinks:* Miles’ Twitter* Benchmark homepage* LangChain homepage Get full access to The Gradient at thegradientpub.substack.com/subscribe

In this episode, Shreya Shankar, a computer scientist pursuing her PhD in databases at UC Berkeley, discusses the challenges faced by machine learning engineers, including ensuring consistent performance and accuracy in production ML models. She also explores the importance of preserving information for anomaly detection, quantitative statistics for data cleaning, and the evolution from MLOps to LLMOps and FMOps. The speakers discuss the difficulties of implementing ML models and ML monitoring, as well as varying perspectives on AI as a product and its impact on customer expectations.

Guest Stevan Harnad, professor of psychology and cognitive science, discusses the Symbol Grounding Problem in AI. They explore category formation, non-linguistic category learning, and the limitations of current AI models. The episode also touches on Alan Turing's motivation, the easy and hard problems of AI, multimodal systems, and categorical perception in machine learning models and humans.

Terry Winograd, Professor Emeritus of Computer Science at Stanford University, shares insights on the atmosphere, conflicts, and collaborations within the MIT AI lab. He discusses the transition from AI to HCI, explores the impact of chatbots on information access and verification, and addresses the challenges of trust in journalism. Winograd also emphasizes the evolving role of HCI in addressing societal issues and the importance of trustworthy AI systems.