Can even artificial intelligence suffer from depression, or experience hallucinations, like the human brain? The question is much less odd than it might seem, according to Zachary Mainen, a neuroscientist at the Champalimaud Centre for the Unknown, a neuroscience and cancer research institute in Lisbon, Portugal, who studies the lesser-known aspects of machines capable of learning.
On the occasion of an international cybernetic symposium in New York, Mainen was interviewed by Science magazine, where he explained that depression and even certain types of hallucinations in humans seem to be due to the altered workings of a molecule called serotonin. This substance, in reality, serves an important role in many other activities which are beneficial to the brain, and therefore there is the possibility of it being “copied” by the artificial intelligence systems used in neuroscience. However, at the same time the “machines” might then experience depression if the computerized equivalent of serotonin should malfunction.
Artificial intelligence, Mainen continued, perceives serotonin and other neurotransmitters, such as dopamine, as “control knobs” and follows similar models. If these knobs get stuck, even artificial intelligence could experience “neurological” problems.
Mainen added that serotonin is strongly implicated in situations in which the world undergoes a sudden and radical change. In these cases, humans which either produce low levels of serotonin, or utilize low levels for other reasons, are not able to adapt in an efficient way, and fall into depression. But even a machine could have similar altered mechanisms, and face frustration, depression and other symptoms, as a consequence of being unable to adapt to unexpected changes in a flexible way.
The science which studies these aspects of artificial intelligence is just at its beginnings, and many studies need to be carried out before the problems and possible remedies can be defined. But at this point, the door has been opened…
