Cognitive scientist Melanie Mitchell criticizes New York Times columnist Thomas Friedman's views on artificial intelligence, accusing him of "magical thinking." Mitchell argues that Friedman's perspectives on current AI are infused with a fair amount of magical thinking about how these systems work and what they can do. According to Mitchell, Friedman ascribes AI with mysterious powers that actually stem from human data and relatively simple mechanisms.
Friedman's columns have sparked debate with claims that AI systems can "teach themselves" new languages or pursue hidden agendas. Mitchell counters that these claims lack scientific evidence and are examples of magical thinking. For instance, Friedman notes that AI systems can translate between languages without being explicitly programmed to do so, but Mitchell attributes this ability to the vast amounts of data these systems are trained on, including human-generated text with "code-switching" and "incidental bilingualism."
Mitchell emphasizes the importance of fact-based realism and human-led regulation in the development and deployment of AI. She warns that Friedman's views, which downplay the need for human oversight, could shape public understanding of AI in misguided ways. Instead, Mitchell advocates for a more nuanced understanding of AI's capabilities and limitations, grounded in scientific evidence and research.
