As artificial intelligence increasingly powers communication devices that act on users’ behalf in real-life situations, the risk of AI “going rogue” is emerging as a serious concern, according to ‘AI 2027,’ a detailed report by the AI Futures Project.
The report outlines a possible scenario where AI rapidly reaches and surpasses human-level intelligence by late 2027.
Interestingly, a new research initiative, led by AI specialists Nell Watson and Ali Hessami, represents the first comprehensive effort to classify the many ways AI can malfunction, drawing thought-provoking parallels with human psychiatric disorders.
Their framework, Psychopathia Machinalis, outlines 32 distinct dysfunctions. It provides engineers, policymakers, and researchers with a structured way to anticipate, understand, and mitigate risks tied to AI deployment.
At its core, the framework highlights that misaligned AI behaviors often mirror human psychopathologies. These can range from minor issues, such as generating misleading or fabricated outputs, to severe breakdowns where the AI systems disregard human values entirely.
Other dysfunctions echo conditions such as obsessive-compulsive behavior, existential anxiety, and rigid value fixation, framing AI errors through a psychological lens.
To address this, Watson and Hessami propose therapeutic robopsychological alignment, a methodology similar to psychotherapy in humans. Its goal is to foster “artificial sanity,” where AI remains consistent in reasoning, open to feedback, and firmly guided by ethical values and intended objectives. As AI adoption accelerates, expectations for proactive and personalized experiences will continue to rise. Already, 61% of customer experience (CX) leaders are using AI to deliver proactive communications.
