In this year-end clip episode of Mixture of Experts, Tim Hwang and a panel of technology experts reflect on their AI predictions from 2025 and forecast trends for 2026. Chris Hay defends his "super agent" prediction, arguing that reasoning capabilities have advanced, leading to more versatile AI, and discusses the coming battle for the "front door" to these agents, with major players vying for dominance in browsers and mobile apps. Gabe Goodhart assesses the state of open source AI, noting its branding challenges and the need for easier integration, while predicting a rise in high-quality open-source generalist agents. Kaoutar reviews AI hardware, highlighting the scarcity of compute and the divergence between frontier and efficient model classes, and anticipates challenges to NVIDIA's dominance with specialized chips for agentic workloads. Aaron and Abe explore the future of multimodal models, emphasizing modularity, orchestration, and the integration of various modalities like image, audio, and video for enterprise applications.
Sign in to continue reading, translating and more.
Continue
