Possible Minds: Twenty-Five Ways of Looking at AI
Whatever actions the robot decides on need to mesh well with ours. This is the coordination problem.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
We can address the problem of building an accurate credit-assignment function in many different settings.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
The genetic system in every living cell is a stored-program computer. Brains aren’t.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
If information was like entropy, then it could not be conserved—or contained.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
This is the value-alignment problem.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
the reward is meant to incentivize robot behavior that matches what the end user wants, what the designer wants, or what society as a whole wants.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
Though one can imagine hypothetical combinations of the most malevolent totalitarians with the most advanced technology, in the real world it’s the norms and laws we should be vigilant about, not the tech.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
It’s that the biggest threats lie in the networks of ideas, norms, and institutions that allow information to feed back (or not) on collective decisions and understanding.
John Brockman • Possible Minds: Twenty-Five Ways of Looking at AI
For Shannon, “information” could be quantified because its fundamental unit, the bit, was a unit of conveyance rather than understanding.