Latest Interviews

Oliver Cameron (CEO, Voyage) - MIT Self-Driving Cars

Oliver Cameron, Lex

Oliver Cameron founded Voyage to deploy Level 4 autonomous vehicles within closed-loop retirement communities, leveraging exclusive licensing agreements to secure defensible market positions while addressing the mobility needs of seniors. Previously accelerating AV talent development through Udacity's program, Cameron applied rigorous engineering solutions like 128-channel LiDAR and deep learning perception networks to eliminate edge cases such as foliage occlusion and pedestrian clustering. The company's strategy prioritizes slow-speed safety and remote human intervention over competing in dense urban centers, aiming to capture a 47-million-person market by integrating dynamic risk assessment with Intact Insurance.

MIT-AVT: Data Collection Device (for Large-Scale Semi-Autonomous Driving)

An MIT-led study utilizes Ryder System's fleet of over 30 vehicles to gather extensive naturalistic driving data, analyzing how humans supervise semi-autonomous systems across more than 320,000 miles. The project employs a specialized hardware architecture to record synchronized video, GPS, and vehicle telemetry with high thermal resilience and precise clock accuracy, generating nearly 300 terabytes of compressed footage for deep learning analysis. Future iterations will transition to NVIDIA Jetson TX2 hardware to enable selective recording of critical edge cases, shifting the research focus toward understanding driver cognitive load and internal behavior.

Emilio Frazzoli, CTO, nuTonomy - MIT Self-Driving Cars

Emilio Frazzoli, Lex

MIT alumnus and Neutonomy CTO Emilio Frazzoli advocates for a direct leap to Level 4 and 5 automation to capture $2 trillion in annual economic value through scalable vehicle-sharing fleets, bypassing the safety risks associated with partial human-supervision levels. While rejecting end-to-end deep learning in favor of formal rule verification via his RRT* algorithm, Neutonomy prioritizes complex urban operations and rigorous mathematical safety theories to address unresolved ethical dilemmas and regulatory gaps. Anticipating rapid adoption of these dedicated autonomous services, the company plans to expand its workforce and scale operations to transform mobility supply within the next two years.

Lisa Feldman Barrett: How the Brain Creates Emotions | MIT Artificial General Intelligence (AGI)

Lisa Feldman Barrett

Neuroscientist Lisa Feldman Barrett challenges the notion of universal, pre-wired emotions by arguing that the brain constructs emotional experiences on the spot to regulate the body's metabolic needs through a process called allostasis. This perspective reveals that emotions are cultural concepts shaped by language and context rather than biological facts, leading to the conclusion that current AI emotion detection is fundamentally flawed because it attempts to read fixed facial signals that do not correspond to intrinsic feelings. Consequently, building truly intelligent artificial systems requires simulating a body with internal regulatory states to generate meaningful affect, shifting the focus from abstract reward functions to the biological imperatives of resource management and social regulation.

Sacha Arnoud, Director of Engineering, Waymo - MIT Self-Driving Cars

Sacha Arnoud, Lex

Following its 2017 spin-off from Google, Waymo has accelerated autonomous driving operations by completing over 4 million miles and launching the first public driverless fleet in Phoenix using custom-equipped Chrysler Pacifica vehicles. The company leverages deep learning and Google's TensorFlow infrastructure to process multimodal sensor data from LiDAR, radar, and cameras, enabling robust perception and planning within a closed-loop system of 25,000 simulated cars. Looking forward, Waymo is expanding its operating domain to complex urban environments like San Francisco while refining its technical architecture to prioritize safety and generalization over memorized scenarios.

MIT AGI: Building machines that see, learn, and think like people (Josh Tenenbaum)

Josh Tenenbaum

Josh Tenenbaum and the Center for Brains, Minds, and Machines argue that current deep learning systems are limited specialized tools that fail to replicate human general intelligence due to a lack of common sense and world modeling. To achieve true Artificial General Intelligence, the proposal advocates for a reverse-engineering approach that integrates cognitive science with engineering to build probabilistic programs capable of "programming" internal models of physics and psychology. This methodology aims to bridge the gap between industry's data-driven pattern recognition and the foundational, low-data learning mechanisms observed in human infants.

MIT 6.S094: Deep Reinforcement Learning

This presentation explores the development of end-to-end reinforcement learning systems that perceive raw sensor data, reason through time, and execute physical actions to achieve complex goals. It details technical innovations like Experience Replay and Target Networks that enabled Deep Q-Networks to master Atari games and AlphaGo Zero to surpass human champions through self-play without human data. Despite these benchmark successes, the discussion concludes that real-world applications in autonomous driving remain limited by data inefficiency, safety challenges, and the unresolved gap between simulated performance and robust physical-world reasoning.

Sertac Karaman (MIT) on Motion Planning in a Complex World - MIT Self-Driving Cars

Sertac Karaman, Sirtesh Karaman, Lex

MIT AeroAstro professor Sirtesh Karaman discusses his pioneering RRT* algorithm, which guarantees optimal trajectory convergence for autonomous vehicles, and reflects on MIT's 2007 DARPA Urban Challenge success where his team developed software now standard in the automotive industry. Karaman outlines current research into ultra-agile robotics and compressed high-dimensional control systems, while detailing the commercial launch of Optimus Ride and projecting the near-term viability of vision-only autonomy and vehicle-to-infrastructure communication networks. The presentation concludes by analyzing how formal optimization and deep learning will transform logistics costs and overcome non-technical regulatory barriers in the evolving landscape of self-driving technology.

Chris Gerdes (Stanford) on Technology, Policy and Vehicle Safety - MIT Self-Driving Cars

Chris Gerdes, Lex

Stanford professor and former USDOT Chief Innovation Officer Chris Gerdes outlines the dual trajectory of autonomous vehicle development, highlighting the high-performance "Shelly" research car's ability to replicate human driving instincts while navigating the constraints of a U.S. regulatory framework based on self-certification. He critiques the lag in formal rulemaking compared to rapid AI advancements, noting that current voluntary guidelines address operational design domains and fallback conditions but struggle to resolve conflicts between rigid traffic codes and safety-critical maneuvers. Gerdes ultimately advocates for data sharing to improve neural networks, the elimination of human error in programming, and a potential redesign of vehicle physics to reduce mass and energy consumption through enhanced safety.

MIT 6.S094: Deep Learning for Human-Centered Semi-Autonomous Vehicles

Researchers are collecting billions of high-speed video frames from semi-autonomous Teslas to train deep learning models that detect critical driver metrics such as body pose, gaze direction, and cognitive load. By shifting from fully supervised to semi-supervised annotation strategies, the team achieves an 84-fold reduction in human effort while accurately identifying micro-saccades and emotional cues to overcome current privacy and trust barriers. This data-driven approach aims to replace static crash test assumptions with dynamic occupant monitoring, ultimately enabling vehicles to adapt passive safety systems based on real-time human behavior.

Foundations and Challenges of Deep Learning (Yoshua Bengio)

Yoshua Bengio, Hugo Larochelle, Andrej Karpathy, Richard Socher, Sherry Moore, Ruslan Salakhutdinov, Andrew Ng, John Schulman, Pascal Lamblin, Adam Coates, Alex Wiltschko, Quoc Le, Shubho Sengupta

Yoshua Bengio outlines five essential ingredients for human-level machine learning, emphasizing that deep neural networks overcome the curse of dimensionality through parallel and sequential composition to efficiently represent complex functions. He contrasts current high-dimensional optimization landscapes, which are dominated by saddle points rather than local minima, against historical theories while highlighting unsupervised learning as a critical mechanism for developing generalizable world models. The presentation concludes by addressing future challenges in training long-term dependencies and integrating neuroscience-inspired alternatives to backpropagation, alongside administrative notes regarding an upcoming textbook by Bengio, Ian Goodfellow, and Aaron Courville.

Torch Tutorial (Alex Wiltschko, Twitter)

Alex Wiltschko, Hugo Larochelle, Andrej Karpathy, Richard Socher, Sherry Moore, Ruslan Salakhutdinov, Andrew Ng, John Schulman, Pascal Lamblin, Adam Coates, Quoc Le, Yoshua Bengio, Shubho Sengupta

This presentation details the practical implementation and theoretical foundations of the Torch deep learning framework using the Lua language, developed in collaboration with experts from Facebook, Google, and Twitter. The speaker explains how Torch leverages LuaJIT for high-performance embedded deployment while utilizing its dynamic Autograd system to support flexible control flow and custom gradients without the overhead of static computation graphs. Case studies from Twitter demonstrate the framework's transition from a research tool for cutting-edge models like GANs to a production environment for serving media, highlighting its efficiency in both training via reverse-mode differentiation and inference through lightweight C++ integration.

Sequence to Sequence Deep Learning (Quoc Le, Google)

Quoc Le, Hugo Larochelle, Andrej Karpathy, Richard Socher, Sherry Moore, Ruslan Salakhutdinov, Andrew Ng, John Schulman, Pascal Lamblin, Adam Coates, Alex Wiltschko, Yoshua Bengio, Shubho Sengupta

This presentation details the evolution of sequence-to-sequence learning for automating email responses, transitioning from bag-of-words models to Recurrent Neural Networks and advanced attention mechanisms. Key technical advancements include the encoder-decoder architecture with beam search decoding, personalized user embeddings, and gated units like LSTMs to manage long-term dependencies and vocabulary limitations. The discussion concludes by highlighting real-world applications in machine translation and conversational AI, alongside future research directions in unsupervised learning and global sequence optimization.

Deep Learning for Natural Language Processing (Richard Socher, Salesforce)

Richard Socher, Hugo Larochelle, Andrej Karpathy, Sherry Moore, Ruslan Salakhutdinov, Andrew Ng, John Schulman, Pascal Lamblin, Adam Coates, Alex Wiltschko, Quoc Le, Yoshua Bengio, Shubho Sengupta

This presentation outlines the evolution of Natural Language Processing from hierarchical linguistic analysis to deep learning architectures like Word2Vec, GRUs, and Dynamic Memory Networks that handle sequence modeling and visual question answering. Key researchers discussed how continuous vector representations and gated units address ambiguity and long-range dependencies, achieving state-of-the-art performance on benchmarks such as Facebook's bAbI dataset and reducing language modeling perplexity to 70. Despite these advances, the discussion highlights persistent challenges regarding unified joint models, data scarcity in specialized domains, and the need for improved robustness against adversarial inputs and false premises.

Deep Reinforcement Learning (John Schulman, OpenAI)

John Schulman, Hugo Larochelle, Andrej Karpathy, Richard Socher, Sherry Moore, Ruslan Salakhutdinov, Andrew Ng, Pascal Lamblin, Adam Coates, Alex Wiltschko, Quoc Le, Yoshua Bengio, Shubho Sengupta

This technical presentation delineates Deep Reinforcement Learning as a sequential decision-making framework that employs neural networks to maximize cumulative rewards through policy gradients and Q-function learning. Key figures in the field, such as those at DeepMind, have leveraged these methods to master complex environments including Atari games, Go, and robotic locomotion by addressing challenges like reward sparsity and non-stationary state dynamics. The discussion further contrasts algorithmic trade-offs between sample efficiency and robustness while outlining future directions like hierarchical structures and model-based approaches to enhance real-world deployment.