AI and the Rise of Techno-Fascism in the United States (Garry Kasparov) (www.theatlantic.com)

Garry Kasparov’s podcast interview with cognitive scientist Gary Marcus frames current AI debate in pragmatic terms: AI is a powerful tool, not an agent of inevitable utopia or doom. Marcus and Kasparov contrast two archetypes of machine intelligence—Deep Blue’s brute-force tree search that explored millions of chess positions versus today’s large language models, which are statistical pattern generators trained on massive text corpora. Marcus emphasizes that LLMs produce an “illusion of intelligence”: they can recite rules (e.g., chess rules) yet still make illegal moves because they lack internal, temporally grounded models of the world and causality. That technical gap explains hallucinations, brittle behavior, and errors even when the underlying facts exist in the training data. The conversation’s significance lies in its policy and safety implications: AI is dual-use, and the main risks come from malicious human actors and accidental harms, not a morally autonomous machine. Marcus and Kasparov call for realistic engineering fixes (better grounding, abstract representations, verification) and political responses (oversight, channeling technology to public good) to prevent misuse that could weaken democratic systems or enable “techno‑fascism.” Their message: treat AI’s limits and governance as central problems—address robustness, accountability, and human control—rather than obsess over speculative sentience.