A computer at the University of Texas at Austin recently claimed responsibility for a terrorist bombing. It had not, in fact, bombed anyone, which is reassuring; less reassuring is the fact that the computer made those claims because it is schizophrenic. Indeed: researchers at the fine institution had induced something resembling schizophrenia in DISCERN, a "neural network... [that] can learn natural languages" and remember and repeat "simple stories," by increasing its rate of learning "so that it did not forget at normal rates." And, lo:
That disruption to the neural network's memory process not only resulted in the terrorist story claim, but also led to the disordered behavior known as "derailment." In those cases, DISCERN responded to requests for specific memories by spewing out a jumble of grammatically correct yet disassociated sentences, making abrupt digressions and leaping back and forth from first- to third-person.
A Yale psychiatrist compared DISCERN's output with the linguistic patterns of schizophrenics and say the tests described above "gave the closest matches." The research may support a theory that schizophrenia results from an ability to forget or filter out irrelevant details, though, as the study author says, it doesn't prove anything.