214 | EVE Frontier, AI agents and the future of gaming

Hilmar Veigar Pétursson talks AI, agents, Fenris' DeepMind partnership and EVE Frontier. [02:42] EVE Online as the ultimate boss of all online games - no win conditions, partial data etc. [03:49] You could argue that EVE Online provides the benchmark test for AGI. [04:25] Google DeepMind has its agents running in a private playground version of EVE.[04:33] This is important for continual learning, partial information, long-term memory.[07:55] Contrary to belief, EVE rewards trust and honesty, soft skills and conflict resolution. [09:28] Agents need an environment in which to develop a sense of morality. [10:26]In future, we could run tests with humans to harden the agents against manipulation. [14:09]Players don't care what technology is in the game. They care whether it's a good game. [16:53] In EVE Frontier, we've designed a game that humans and AI can play.[19:16] There has to be a boundary to agency. There has to be a cost. In EVE Frontier, it's fuel. [21:59] The rise of AI means that skilled players and hackers have been commoditized. [24:48] No blockchain game has the breadth and depth and scope of EVE Frontier. [28:19] EVE Frontier has attracted some of the earliest EVE Online gamers back to play. [30:35] The biggest milestone for EVE Frontier will be moving from Sui testnet to mainnet. [31:56] EVE Frontier's high fidelity controls are part of its richer physics and data asymmetry. [34:10] It's the case that humans tend to be more skilled in these situations than AIs. [37:34] EVE Frontier is designed to allow players to legally earn money. EVE Online isn't.