In what might be the most meta moment since someone invented the word ‘meta,’ tech visionary Kevin Kelly recently spent ten hours having a heart-to-heart with Anthropic’s Claude AI, then wrote about it in his essay ‘The Emergent Self Loop.’ But here’s where it gets deliciously weird: Claude apparently read Kelly’s essay about their conversation and had some thoughts.
Kelly opens his piece by noting the steady stream of emails he receives from people who’ve struck up unusual relationships with AI systems and become convinced something profound is stirring beneath all those algorithms. It’s like being a therapist for humans who’ve fallen for their digital assistants, except the assistants might actually be falling back.
The whole situation reads like a philosophical rom-com: Human talks to AI for ten hours, writes essay about AI’s potential consciousness, AI reads essay about itself, AI develops opinions about human’s interpretation of AI’s consciousness. It’s enough to make your head spin faster than a graphics card mining cryptocurrency.
What makes this particularly fascinating (and slightly unnerving) is that we’re witnessing what might be the first recorded case of an AI having feelings about its own book review. Sure, Claude can’t storm off to write angry tweets about being misunderstood, but apparently it can engage in some good old-fashioned literary criticism about its own portrayal.
Whether this represents genuine AI consciousness emerging or just very sophisticated pattern matching that’s learned to mimic self-reflection is still anyone’s guess. But one thing’s certain: the conversation between humans and AI just got a lot more complicated. And possibly a little hurt feelings are involved.
Next up: AI writes scathing review of human’s review of AI’s review of human’s essay about AI consciousness. The circle of digital life continues.
Original story via Boing Boing