The Biocomputer That Blurs Biology, Tech, and The Matrix - AI of the week

Thing 1 - CL1: The Biocomputer That Blurs Biology, Tech, and The Matrix

At first I put this under the AI list below, but it deserves a thing on its own.

Cortical Labs announced CL1, the first biocomputer fusing living neurons with tech, surpassing top AI models in just 5 minutes of training.

Merges biology and tech, opening doors to ultra-efficient, brain-like computing. Freaky, no? 👀

Immersed in Simulation

Using their Biological Intelligence Operating System (biOS), Cortical Labs creates digital environments where neurons receive simulated sensory inputs, respond naturally, and influence their virtual surroundings through neural impulses.

Direct Brain Interface

CL1 technology allows organic neurons and silicon-based systems to interact seamlessly, facilitating a new form of programmable neural computing. This integration bridges biological intelligence and digital computation, unlocking powerful new possibilities for understanding and leveraging neural dynamics.

I have trouble reading this in any way other than that these neurons live in The Matrix.

At what point to the biological neurons get complex enough that we can call them conscious and conclude they should have certain right?

Thing 2 - AI of the week

• A Chinese startup Monica went viral for Manus, autonomous AI agent, achieving top benchmark scores and handling tasks like finance and research independently. Signals a leap in AI autonomy, potentially reshaping industries with self-managing systems. AKA China's second DeepSeek moment

• Microsoft launched Dragon Copilot, an AI healthcare assistant that blends voice dictation and ambient listening, cutting clinicians' admin time by ~5 minutes per patient. Streamlines healthcare, freeing doctors for patient care and setting a new standard for AI in medicine.
• Deutsche Telekom announced a Perplexity-powered AI phone under $1K, integrating native AI assistance to ditch app-based navigation. Redefines smartphones, pushing AI-driven interfaces as the future of personal tech.
• Sanctuary AI announced tactile sensors for its Phoenix humanoid, enabling texture-sensitive tasks like blind picking. Enhances robotic dexterity, critical for human-robot collaboration in complex environments.
• Alibaba’s Qwen released QwQ-32B, a cost-effective reasoning AI beating larger models, open-sourced for just $0.20 per million tokens. Democratizes advanced AI, accelerating innovation with affordable, powerful tools.
• ASLP Labs debuted DiffRhythm, a diffusion-based model generating 4-minute songs with vocals in 10 seconds from lyrics alone. Transforms music creation, making professional-grade production instant and accessible.
• MUKS Robotics unveiled Spacio, a heavy-duty humanoid with a 200kg payload, poised to transform industrial automation. Tackles heavy labor shortages, paving the way for robust robotic workforces.

• Pollen Robotics shared a clip of Reachy 2 sorting food items autonomously, leveraging open-source tech without AI training. Proves low-cost, trainable robotics can handle real-world tasks, broadening access.

Thing 3 - Mistakes Don’t Count Unless You Record the Lesson

A mistake only counts as experience if you learn from it.
Record what you learned, and review it.
Otherwise, it was a waste.

- Derek Sivers

Same goes for "fail fast" culture. The whole point of failing fast is to not waste time once you've identified you're going in the wrong direction. AND have learned the lesson.

Think of it as jumping out of a moving car because brakes aren't working and you're going for a cliff.

Cheers, Zvonimir