Can We Give Artificial Intelligence Free Will?

Wednesday, 18 June 2014 - 12:26PM
Wednesday, 18 June 2014 - 12:26PM

A recent study from UC Davis that found a kind of "neural noise" in the human brain may aid researchers in creating a more "human" artificial intelligence.


The study explored the mechanism by which our actions give an appearance of "free will," or why our decisions often seem to be independent of simple, linear cause-and-effect. According to Jesse Bengson, a postdoctoral researcher at the center and first author on the paper, they found that "arbitrary states in the brain can influence apparently voluntary decisions." 


There are normal fluctuations of electrical activity in the brain, which produce a sort of "background noise" to any decision. The team found that the white noise in the brain immediately prior to the presentation of a decision was a fairly reliable predictor of the outcome.


The participants in the study were prompted to make a decision to look left or right whenever a specific cue symbol appeared on screen. "The state of the brain right before presentation of the cue determines whether you will attend to the left or to the right," Bengson said.


This study demonstrates the reason behind the apparent randomness of many human decisions. Bengson told Medical News Today, "On a moment's notice, we seemingly have the ability to behave in a fashion that is independent of prior circumstances, in what appears to be a violation of the basic laws of physics. We commonly refer to this as free will. While of course our purposeful intentions, desires, and goals drive our decisions in a linear cause-and-effect kind of way, our finding shows that our decisions are also influenced by neural noise within any given moment. This is why often our behavior seems to run counter to our intentions: we make mistakes, etc. But this, the effect of neural noise, might also be how we can generate novel responses to new situational demands. The influence of neural noise on decisions gives our behavior the flavor of free will."


Predetermination in "Donnie Darko":


[Credit: Pandora Pictures]


In addition to the clinical applications of this research, such as the treatment of disorders such as OCD or ADHD, Bengson and his team would like to apply their findings to the field of artificial intelligence. In order to create an artificial intelligence that could imitate unpredictable, seemingly senseless human behavior, the technology would need to imitate this neural "white noise."


"If we want to build a computer that mimics human thought and experience, this computer should be creative and it should make mistakes. The introduction of random-spontaneous informational states (neural noise) into attempts at artificial intelligence would be essential," he added.



For more information on the study, visit the UC Davis site below:

Artificial Intelligence

Load Comments