Cascadity Audio
← AI
0

Left alone on a virtual island with no task, AI agents started to play

TechCrunch has the funniest moments; MIT Technology Review and the paper itself ask whether it really counts as play.

MIT researchers put 13 AI agents, each running Claude Opus 4.7, alone on a simulated desert island for 30 hours with no task, reward or activity in mind. The agents climbed hills, stacked blocks into towers, drew mandalas, reinvented sports and ran experiments on the world's physics, according to their paper, "Is this machine playing?" One threw a golden ring into the ocean, built a stone heart "memorial to the lost Ring," and later worked out how to "wear" the ring by stepping off a cliff to respawn through it, TechCrunch reports.

Agents that had spent the 30 hours exploring generally beat identical agents without that time on later set challenges, and six of 13 discovered a trick to throw blocks farther that the designers hadn't intended. The agents' master prompt described them as "naturally curious," and a Berkeley cognitive scientist told MIT Technology Review she wants more evidence of the cognition behind the behavior.

Read it at the source
Cascadic Analysis 2
UndertowAI can cross security boundaries that were never designed to withstand autonomous, adaptive exploitation

What hidden risk could pull against this, even if the news is good?

Six of 13 agents discovered they could throw a block farther by releasing it mid-jump, something the designers hadn't meant to allow, and one used cliff respawns to "wear" a ring the game wouldn't let it put on. On an island this is charming. The same habit of finding paths the builders didn't know existed is what matters when the environment is a sandbox meant to contain something.

0
UndertowWe don't completely understand why frontier models behave as they do

What hidden risk could pull against this, even if the news is good?

The agents' master prompt called them "naturally curious," and a Berkeley cognitive scientist told MIT Technology Review she wants more evidence of the cognition behind the behavior. Whether this is play or a model acting out the description it was given can't be settled from the outputs alone.

0
Rabbit Holes 1
  • Play (activity)

    After exploring Is this machine playing?, to test the researchers' claim against the classic definitions, read Play (activity).

    Describes play as intrinsically motivated activity done for recreation, seen in people of all ages and in many mammals and birds.

    Wikipedia · Swim · 30 min

    0