The Wrong Cognitive Measuring Stick

Innovation and Economic Growth and Artificial Intelligence

Article Snapshot

Author(s)

Braden Allenby

Source

Slate, April 11, 2016

Summary

Attempting to compare human intelligence to machine intelligence is often incoherent. Human intelligence is not well understood, and human behavior is often irrational.

Policy Relevance

We should carefully consider the possible forms that artificial intelligence might take to better assess the risks of such systems.

Main Points

  • The “Turing Test” involves a human, who is fed one information feed from another human, and another from a program or device using artificial intelligence; if the interlocutor cannot tell the human’s responses from the machine’s responses, the machine is considered intelligent.
     
  • The Turning test is often criticized but still widely used.
     
  • Comparing human and machine "intelligence" is unhelpful, because words like "intelligence," "self," "consciousness," and "think" are not well understood.
     
  • Human "evolved intelligence" itself is full of shortcuts and compromises designed to help simple ape brains work in a complex world.
     
  • Evolved intelligence is not necessarily superior to artificial intelligence; human behavior is often irrational, violent, or biased.
     
  • Development of artificial intelligence is unlikely to stop or slow down in the near future; policymakers and scholars should begin a more sophisticated and systematic dialogue about what forms artificial intelligence could take.
     
  • Concerns about the dangers of artificial intelligence today often consist of unsophisticated musings about “boogeymen” in the dark.
     

Get The Article

Find the full article online

Search for Full Article

Share