The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)

Dillon Erb, CEO of Paperspace, dives into the world of compositional machine learning and its potential to revolutionize software development. He discusses the evolution from experimental ML to a disciplined engineering framework and the challenges of integrating cloud technology. The conversation highlights the debate around the role of notebooks versus traditional coding practices in data science. Also on the table is Paperspace’s innovative new product, Workflows, which aims to streamline ML application development into a more cohesive system.

Yanshuai Cao, a Senior Research Team Lead at Borealis AI, discusses his groundbreaking work on Turing, an engine transforming natural language into SQL queries. He compares it with OpenAI's Codex, highlighting the unique challenges of SQL generation. The conversation reveals insights into the crucial role of reasoning and common sense in accurate query creation. They also tackle complexities in multilingual datasets, data augmentation, and the ongoing quest for model explainability, shedding light on fascinating advancements in AI technology.

Yejin Choi, a professor at the University of Washington, dives into the fascinating world of common sense reasoning and its role in natural language processing. She shares insights from her recent keynote at the Stanford HAI workshop, discussing how AI can enhance creative storytelling and the challenges in achieving human-like reasoning. Topics include the evolution of knowledge graphs, the intersection of common sense and narrative, and the importance of integrating knowledge in AI. Yejin also challenges conventional views on AI scalability in the quest for Artificial General Intelligence.

In this discussion, Konrad Tollmar, research director at Electronic Arts and an associate professor at KTH, dives into the groundbreaking work of EA's SEED team. He reveals how deep reinforcement learning is transforming game testing, focusing on titles like Apex Legends and FIFA. Konrad highlights innovative methods for glitch detection using CNNs and the challenges of training AI in complex 3D environments. He also shares his vision for the future of AI in gaming, emphasizing the importance of adaptability for enhancing player experiences.

Kai-Fu Lee, Chairman and CEO of Sinovation Ventures and author of AI 2041, discusses the future of AI through intriguing 'scientific fiction' stories. He reveals why he chose a 20-year horizon for these tales and elaborates on the implications of level 5 autonomous driving, job displacement, and how AI might reshape work. Kai-Fu also tackles ethical concerns around autonomous weapons and the regulatory landscape, urging listeners to be proactive about AI's potential risks and benefits.

Daniela Rus, Director of CSAIL and Deputy Dean of Research at MIT, shares her fascinating journey in robotics and AI. She offers insights into the history and impact of CSAIL, emphasizing the importance of physicality in robotics. Rus discusses her innovative work in soft robotics and the creation of a mini surgeon robot, as well as advancements in autonomous vehicles. The integration of AI and robotics is explored, highlighting efficiency and safety in machine learning, showcasing their potential to transform various fields.

In this discussion, Alexander Richard, a research scientist at Facebook Reality Labs and ICLR Best Paper Award winner, shares insights into his groundbreaking work on binaural audio synthesis. He dives into the challenges of audio representation in noisy environments and the complex process of generating realistic spatial audio from mono sources. Richard also highlights the difficulties of dynamic time warping and the need for accurate 3D measurements in virtual reality. His thoughts on Codec Avatars and future research directions promise to reshape how we experience sound and presence in virtual spaces.

Today we’re joined by Alona Fyshe, an assistant professor at the University of Alberta. We caught up with Alona on the heels of an interesting panel discussion that she participated in, centered around improving AI systems using research about brain activity. In our conversation, we explore the multiple types of brain images that are used in this research, what representations look like in these images, and how we can improve language models without knowing explicitly how the brain understands the language. We also discuss similar experiments that have incorporated vision, the relationship between computer vision models and the representations that language models create, and future projects like applying a reinforcement learning framework to improve language generation.The complete show notes for this episode can be found at twimlai.com/go/513.

In this engaging conversation, Samory Kpotufe, an associate professor at Columbia University, delves into the complexities of adaptive algorithms in machine learning. He highlights the importance of self-tuning algorithms that can adjust to varying data. The discussion covers transfer learning, emphasizing practical applications and challenges. Samory also touches on innovative methods in unsupervised learning and anomaly detection, especially within resource-constrained devices. His insights into the intersection of fractals and high-dimensional data add a fascinating layer to the conversation.

Eric Rice, an associate professor at USC and co-director of the USC Center for Artificial Intelligence in Society, sheds light on the intersection of AI and social science. He shares his experiences working on projects like HIV prevention for homeless youth and using machine learning to aid in housing resource allocation. Eric emphasizes the need for interdisciplinary collaboration and discusses how social scientists approach assessment differently than computer scientists, focusing on real-world impacts of AI solutions.