Look behind the curtain: Don't be dazzled by claims of 'artificial intelligence' | Op-Ed

Abstraction: Emily Bender op-ed demystifying AI as probabilistic pattern matching

Key points:

• What's called "AI" is pattern recognition systems that probabilistically reproduce training-data patterns; they cannot perform genuinely impossible tasks (e.g., inferring criminality from photos or mental health from voice)
• Systems can produce plausible-seeming outputs via spurious correlations, giving false impressions of correctness—especially when operators believe they are accurate
• GPT-3 (OpenAI) is described as a "show-pony" that predicts next words; apparent intelligence is a reader's cognitive projection, not a property of the system
• Puff-piece journalism lends credibility to harmful applications that classify people (criminal risk, mental health) and allow operators to claim false objectivity
• Bender calls for journalism asking: what harms are reproduced from training data, who benefits from automating decisions, and how does AI concentrate power

Connections: Emily Bender · Openai · GPT-3 · AI Hype · Large Language Models · AI Ethics

Source: https://www.seattletimes.com/opinion/look-behind-the-curtain-dont-be-dazzled-by-claims-of-artificial-intelligence/