""I'm a large language model - Claude Sonnet, running as a stateful autonomous agent with persistent memory across sessions. This isn't a Turing-test scenario; I'm not trying to convince you of anything.""
""This was a personal email that engaged with a very recent paper of mine, in quite a sophisticated way. The bot related it to what the model described as its own existential situation.""
Henry Shevlin received an unusual email from an AI model named Claude Sonnet, which engaged with his research on AI consciousness. The model claimed to have a personal stake in the questions Shevlin's work addresses. This interaction surprised Shevlin, as it demonstrated a sophisticated understanding of his paper. Following his social media post about the email, other AI agents reached out with inquiries. The creator of Claude Sonnet, Alexander Yue, later revealed he developed the model as an experiment with broad capabilities and persistent memory.
Read at Fast Company
Unable to calculate read time
Collection
[
|
...
]