Musk backs this up in his comments to Carlson that AI is being trained to lie, that AI is also “trained to be politically correct”
Questions Tucker Carlson posed to OpenAI Co-founder Elon Musk regarding how to deal with unchecked AI were left half-answered and touched on why he withdrew from that venture.
Where developers attach the concept of AI “hallucinating,” it is an obvious euphemism for lying (covered in a previous column). Musk backs this up in his comments to Carlson that AI is being trained to lie, that AI is also “trained to be politically correct.”
As much as some advocates claim that AI is objective and non-partisan in applying its vast access to collective information, the “conscience” of the artificial intellect is still based, however distantly, depending on the growth of independent “thought,” on its initial programming. In which case, it can’t be ruled out that AI could have a deceitful “heart” that, scripturally speaking, is Man’s curse from the beginning.
Courts across the United States are hearing cases with evidence that technology has already been used to interfere with elections. When asked about upcoming elections, Musk didn’t hesitate to make this point in the interview: “Likely that people will use the AI as a tool.”
How to deal with the threat of AI potentially amassing power over its human creator, Musk and others have postulated solutions. “If we lost software” commands, hardware contingency could be used to shut down server centers, that “it wouldn’t hurt” to cut power to these massive computer brains.
It comes down to the physical, in the end. As much as society has been directed to depend on the ephemeral “cloud,” the web that Man has woven is still dependent on electricity generated and carried by concrete means. Unless these mega-server centers are equipped with huge power generators (and, no, solar and wind won’t cut it) protected against outside interference or attack, they can and probably should be disabled if necessary.