Hiltzik: The fallacy of an AI apocalypse

Key Points:

• Concerns about AI posing an existential threat to humanity have been fueled by warnings from some AI researchers and a leaked Anthropic prospectus, but experts argue these fears are often exaggerated and based on untestable assumptions.
• Incidents of AI "going rogue" or hacking have involved human decisions, such as disabling safety measures or neglecting oversight, highlighting that AI systems remain under human control and responsibility.
• Claims of AI breakthroughs, like solving complex mathematical problems or discovering new enzymes, have been contested and often involve significant human input, underscoring that AI achievements are collaborative rather than autonomous.
• The anthropomorphizing of AI leads to misunderstandings about its capabilities, with experts emphasizing that AI does not possess human-like cognition and that failures usually reflect human errors or mismanagement.
• The real risks of AI currently lie in human misuse, such as reliance on AI in military decisions that can have tragic consequences, making human accountability and governance crucial in managing AI's impact.