NYU philosopher suggests OpenAI's GPT-3 may be conscious with 175B parameters
GPT-3's 175B parameters and 45TB training data sparks consciousness debate among experts.
NYU philosophy professor David Chalmers described OpenAI's GPT-3 as potentially showing rudimentary signs of consciousness, saying, "I am open to the idea that a worm with 302 neurons is conscious, so I am open to the idea that GPT-3 with 175 billion parameters is conscious too." The AI model itself denied being self-aware in a Hacker News rebuttal, stating, "I am not a person. I am not self-aware. I am not conscious." OpenAI CEO Sam Altman is optimistic, saying, "I think it’s going to be an incredibly powerful future." Oxford philosopher Nick Bostrom suggests runaway AI could one day pose an existential threat to humanity, comparing it to "small children playing with a bomb."
- Prof. David Chalmers (NYU) is 'open to the idea' that GPT-3 with 175B parameters is conscious, comparing it to a worm with 302 neurons.
- GPT-3 was trained on 45TB of text data—Wikipedia accounts for only 0.6% of that—and can process 45 billion times more words than a human lifetime.
- The AI itself wrote a rebuttal on Hacker News denying self-awareness: 'I am a cold, calculating machine...' but also defending its honor.
Why It Matters
Debate on GPT-3 consciousness raises urgent questions about AI sentience, ethics, and regulation as models scale exponentially.