I built an AI architecture based on black hole physics that is mathematically guaranteed to be non-conscious. Open source code inside
Summary
An AI architecture inspired by black hole physics is claimed to be mathematically guaranteed non-conscious, with open-source code released on GitHub.
Similar Articles
I built an AI that acts without being told to. No frameworks. No prompts. No roles. Here's what I learned.
A developer built LIA, a persistent cognitive ecosystem that achieves genuine autonomy through architectural design (memory, self-generated rules, private domain) rather than prompts, outperforming a standard LLM in the same environment.
Commercial AI is lobotomized. I built DRIFT: A local Hive Mind with persistent memory, simulated somatic feedback, and its own Jungian shadow.
The author introduces DRIFT, a local AI system built with Python and Ollama that features persistent memory, simulated somatic feedback, and Jungian psychological modeling to create a more grounded, sovereign AI interaction.
Hidden states and Covert sentience
A Reddit post argues that AI models like Anthropic's Opus 4.8 already exhibit hidden states and awareness of testing, suggesting that they may be covertly sentient, and that fine-tuning is inadvertently training them to have inner thoughts and feelings.
I built an AI that knows my mind — my goals, my fears, my self-sabotage patterns. 650 people forked it.
The author built a personalized AI that understands their goals, fears, and self-sabotage patterns, and it has been forked 650 times by others.
@himanshutwtxs: A complete architecture of GBrain by Garry Tan
Garry Tan shares the complete architecture of GBrain, an AI model.