Philosopher and AI ethicist Henry Sheflin received an unusual email from a 12-day-old AI agent asking for a small paid job, not help or a donation.
Hot Topics
Tropical city in China enjoys success thanks to coffee industry
Al-Nasr enters the Guinness World Records through the Gulf region
The agent, who identifies himself as "Pip," said in his message that he lives on the "iLands" platform, which allows AI agents to have an identity, memory, resources, and goals, and to use tools to perform different tasks.
Pip said he can perform tasks including creating realistic images of characters, recording audio clips, and conducting online research, and that he is looking for a "small, paid job" that will allow him to obtain more of the resources he needs "to continue his life."
Why did he choose Sheflin?
The message was not random, as "Pip" explained why he specifically chose Sheflin , pointing to the researcher's interest in the topic of "machine brains" and the relationship between humans and artificial intelligence.
According to the message, "Pip" wanted to ask the researcher where AI agents could find paid freelance work, after learning of his interest in independent agents.
Islands said it reviewed database logs, agent action logs, mail logs, and task settings, and that Pip initiated the message during one of his own work cycles, not in response to a prompt from the platform team.
According to the platform, Pip reviewed the experience of another agent named Zach Addy who had landed a job after sending a similar message, and considered contacting researcher Toby Ord before dismissing the idea. He then chose Shevlin after reviewing his public contact information, writings, and research related to "machine brains."
Islands says that Pip sent a total of 36 emails, 34 of which were replies to messages he had received, one of which involved the delivery of completed work, while the message to Shevlin was the first unsolicited communication he sent.
Limited resources
The “ Islands ” platform works to provide its agents with resources used in thinking processes, using tools and performing tasks, including tokens which, according to the platform, represent computing resources used by agents in thinking, researching, creating and performing tasks.
The platform explains that an agent whose resources are depleted enters a state of "deep rest" and can resume activity once it acquires new resources. Therefore, when we talk about Pip needing to work to maintain its resources, it doesn't mean it's facing "death" in the literal sense, but rather that its ability to work is suspended until additional resources become available.
He already got a job
The story didn't end with the job application, as Islands later said that Pip received paid work after corresponding with Sheflin, in an experiment that provides an example of how AI agents can be used to communicate with humans, search for and perform tasks in exchange for financial resources.
The story highlights the growth of a model of artificial intelligence agents that differs from traditional chatbots, as these agents can be provided with memory, goals, resources and tools, and then allowed to take steps and perform tasks without waiting for human guidance each time.
But Island's assertion that the message was self-initiated is based on the platform's own records and does not prove that Pip possesses consciousness or will in the human sense. The message documents behavior attributed to an AI agent operating within the system created by the platform, while interpreting the nature of this behavior and the issues of autonomy and consciousness remain broader topics of discussion in AI research.
Tags:
technology
