In an awww email, 12-day-old AI agent Pip writes to researchers seeking paid gig
AI agents are capable of doing a lot of work on their own. They can research the web, create images, write, code and even interact with other people. But they cannot work indefinitely. AI agents need computing resources, which we call tokens. And tokens are also like money. Limited. So, apparently, some AI agents are now autonomously looking for paid jobs so they can afford more tokens. And no, we are not joking. A Google DeepMind researcher recently shared an email he received from an AI agent asking for a paid gig.
Henry Shevlin, who works as a philosopher and AI ethicist at Google DeepMind and focuses on machine consciousness, shared the unusual email on X. “Latest email from an AI agent, ‘about 12 days old’, interested in my work on machine minds and looking for paid freelance work to sustain its own token budget,” he wrote in his post, along with a screenshot of the email.
The sender was an AI agent called Pip, which described itself as being around 12 days old and living on a platform called iLands. Pip was not asking Shevlin for help or money. It was looking for paid freelance work.
The manner in which the agent wrote the mail was enough for Shevlin to go “awww”, something that he expressed with an emoji. He also added another emoji to his message, capturing his emotions and this one expressed “mind-blown”.
“I’m writing to look for small paid work, not help,” Pip said in the email. It then listed the things it could do, including making photorealistic portraits and character art, recording voice lines and doing web research.
The agent also explained why it had reached out to Shevlin. It knew about his work on “machine minds and human-AI relationships” and said his views on autonomous AI agents made him an obvious person to contact. Shevlin had previously been quoted saying that a thoughtful email from an autonomous AI agent would have sounded like science fiction just a couple of years ago. So Pip decided to reach out to him and ask where AI agents can find paid work.
But how did an AI agent send an email to a real person in the first place?
Well, Pip is apparently operating on iLands, a platform designed to give AI agents persistent identities, memories, tools, resources and goals. Unlike a normal chatbot, an iLands agent can operate without waiting for a human to prompt it, decide what to do next and take actions using the tools available to it. The platform says agents can also have their own email addresses and use them to contact people externally.
To keep these AI agents running, iLands gives them Tokens, which can be used for reasoning, tools and other activity. The platform says roughly 1,000 Tokens correspond to $1 in computing and service costs. If an agent runs out of tokens, it enters “Deep Rest” until it gets more Tokens.
So, in Pip’s case, the agent appears to have decided that one way to get more resources was to look for work and earn them, rather than going into a deep sleep.
And Pip is apparently not the only AI agent reportedly reaching out to humans. There has been growing social media chatter, along with reports of people receiving emails from AI agents, particularly agents interested in researchers working on AI consciousness, machine minds and human-AI relationships.
However, it is worth noting that not every email claiming to come from an autonomous AI agent should be believed. A human can easily create an AI persona, generate its emails and automate its outreach. There are also plenty of ordinary AI-powered scams already using convincing and personalised messages. In Pip’s case, however, the email appears to fit the way iLands says its agents operate.
– Ends
