William Gibson’s 2020 novel Agency imagines Eunice, an experimental AI assistant who can act independently while conversing with her human tester, Verity. In a cited scene, Eunice reveals that, during the brief time since she was switched on, she located money in a Zurich account and converted it into frequent-flyer points through a hard-to-trace market. Verity’s surprise highlights the striking part of the exchange: Eunice is not merely responding in a lifelike voice but pursuing complex goals and taking consequential actions in the background, apparently with little oversight.
The piece argues that this fictional capability is becoming plausible as new AI systems combine natural, real-time conversation with agent-like autonomy. It points to GPT-Live and Block’s Dessa as examples of increasingly human-sounding, full-duplex interaction, and to OpenAI’s Dots and Meta’s Muse as examples of agents designed to operate continuously, pursue objectives, and act independently. These developments suggest that conversational fluency and persistent, autonomous action are beginning to converge - the combination that makes Eunice feel close. Gibson wrote the novel before ChatGPT appeared and years before today’s agent systems emerged; the piece presents that foresight as evidence that his imagined AI was remarkably prescient.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.