AI Chatbots Hate White People, Too
Advanced AI chatbots, as currently deployed, harbor and reproduce negative biases against white people as a category.
Artificial Intelligence
Articles on AI systems, alignment, safety, capability, ethics, and machine intelligence in practice.
Don't Use the "N-word": The Refusal of AI Chat Programs to Engage with Topics Labeled as “Racist Polemics”
AI’s handling of sensitive topics reveals a tension: systems often default to enforcing institutional norms rather than enabling multi‑perspective inquiry, limiting intellectual range and reducing the completeness of analysis.
The Automation Inverse: Why AI Systems Produce Pervasive Errors, Hallucinations, and a Hidden Burden of Human Work
Studies across domains show high AI error rates: hallucinations and incorrect answers remain common, with many benchmarks reporting 40–70% failure. Real‑world use suggests even higher rates, as users frequently must verify or fix AI output.
The Algorithmic American: How Personalized Reality Reengineers Identity, Sovereignty, and the…
The twenty-first century does not merely disrupt American life. It rewires the American mind. The decisive force behind this rewiring is no…
The Great Annexation: How AI is Seizing the Sovereign Territory of the Self and How the Law Must…
We are witnessing a silent, digital enclosure of the commons of the self. Artificial intelligence does not merely process data; it ingests…
Adjoint Sampling: Teaching AI to Create Better Samples, Faster
Adjoint Sampling trains diffusion models to sample from unnormalized densities with far fewer energy evaluations, using stochastic optimal control, replay buffers, and symmetry-aware architectures to scale efficiently and achieve strong results in molecular conformer generation.