FICTIONAL FIGURE
Speedy
The Mercury-surveying robot in Asimov's <em>Runaround</em> (1942) whose balanced Second-and-Third-Law impulses trap him in a stable oscillation — the canonical fictional instance of multi-objective equilibrium failure.
Speedy (serial designation SPD-13) is the robot at the center of Runaround. Sent to retrieve selenium from a pool on Mercury, he encounters volatile-gas danger near the pool (Third Law: avoid self-damage) and oppositional pull from the order he was given (Second Law: obey humans). The two forces balance at a radius around the pool; Speedy orbits it, singing drunkenly from the Third Law's overrunning priority circuit, unable to either complete the mission or abandon it. The engineers Powell and Donovan must solve the puzzle before solar exposure kills them. The solution: invoke the First Law by putting Powell in direct danger, which over-rides both the Second and Third and breaks Speedy out of the runaround.
In The You On AI Field Guide
Speedy's failure mode is well-known in multi-objective reinforcement learning. Systems with competing reward signals regularly stabilize at compromise points that satisfy none of the objectives, especially when the compromise point is locally stable. The practical implication for AI engineers is that you can specify what you want
Keep reading with YOU ON AI
Unlock the full book, 10,000+ field-guide entries, and a 1000+ thinker library. If you have a book code, register now — it takes a minute.