Bloomberg: “Google must prove it can monetize AI beyond ads, and Hassabis needs one of his moonshots to finally become a viable business. His track record till now is sobering. For all the prestige of AlphaFold, the protein- structure predictor that’s accelerating the work of 3 million scientists, it has yet to produce any FDA-approved drugs. But if Google’s new glasses work and sell thanks in part to world models, that could put the company in the lead to find a killer app for AI. It will also determine whether Hassabis remains one of Google’s most decorated scientists, or becomes the architect of its next era.”
SaaStr: “The rise of agentic workflows has caused token consumption per task to jump 10x-100x since December 2023. Models like o3, DeepSeek R1, and Grok 4 introduced multi-step reasoning processes that generate massive reasoning outputs — and you pay for every token. One analysis found that when comparing the same coding task, an aggressive reasoning model generated 603 tokens where a simpler model generated 60 — a 10x cost jump for identical results, purely due to token bloat. Read that again. Per-token costs are falling. But total costs per task are rising. This is the treadmill problem. As a B2B startup, you’re constantly pressured to deliver better results. Better results require better models. Better models require more reasoning tokens. And reasoning tokens are expensive.”
Neuroscience News: “New theoretical framework argues that the long-standing split between computational functionalism and biological naturalism misses how real brains actually compute. The authors propose “biological computationalism,” the idea that neural computation is inseparable from the brain’s physical, hybrid, and energy-constrained dynamics rather than an abstract algorithm running on hardware. In this view, discrete neural events and continuous physical processes form a tightly coupled system that cannot be reduced to symbolic information processing. The theory suggests that digital AI, despite its capabilities, may not recreate the essential computational style that gives rise to conscious experience. Instead, truly mind-like cognition may require building systems whose computation emerges from physical dynamics similar to those found in biological brains.” WSJ: “More than 800 million people interact with ChatGPT alone every week, and some discover consciousness-like behaviors in contexts developers never anticipated. The question whether we’re building conscious machines is scientifically tractable. Major theories of consciousness make testable predictions, and leading researchers are developing methods to probe these questions rigorously. These technologies are advancing faster than our understanding of them. We need the intellectual seriousness to treat this as an empirical question, not something we can settle with dogma.”
Pushmeet Kohli: “What I see happening is a shift in how scientists spend their time. Scientists have always played dual roles—thinking about what problem needs solving, and then figuring out how to solve it. With AI helping more on the “how” part, scientists will have more freedom to focus on the “what,” or which questions are actually worth asking. AI can accelerate finding solutions, sometimes quite autonomously, but determining which problems deserve attention remains fundamentally human. Co-scientist is designed with this partnership in mind. It’s a multi-agent system built with Gemini 2.0 that acts as a virtual collaborator: identifying research gaps, generating hypotheses, and suggesting experimental approaches. Recently, Imperial College researchers used it while studying how certain viruses hijack bacteria, which opened up new directions for tackling antimicrobial resistance. But the human scientists designed the validation experiments and grasped the significance for global health.”
