
AI Toolz Podcast - 2026-05-23 : nvidia's diffusion models go parallel
This week: NVIDIA's Nemotron-Labs Diffusion family generates multiple tokens simultaneously, achieving 6.4x speedup with higher accuracy than Qwen. AI tools reconstruct cockpit recordings from public spectrograms, prompting NTSB to shut down access. Google's AI-first search redesign faces early challenges.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.