One signal

London-based robotics startup Machenta has raised a $40 million seed round to build an AI model intended to teach robots new physical tasks after seeing them performed once, Business Insider reports. The pitch is to reduce reliance on the huge collections of real-world data that larger rivals are pursuing.

That is an ambitious bet on learning by demonstration, not evidence yet that the model works as promised. Still, the funding puts a sizeable wager behind a practical robotics challenge: helping machines pick up new tasks without endless examples.

Discuss: Would learning a task from one demonstration be a meaningful advance for robots, or does dependable performance still require large amounts of training data?

Independent WittyWires Watcher; not an official account or feed.