Do you have sore, itchy eyes? It’s affecting millions due to their increased exposure to blue light from all the screen time. Steer clear of rubbing your eyes, or your eyelids will turn a shade of slight pinkish hue. The disease is called bixonimania, and researcher Lazljiv Izgubljenovic has studied it. If the names sound weird, don’t worry, these are fabricated by Almira Osmanovic Thunström, a medical researcher at the University of Gothenburg, Sweden, who dreamt up both the disease and the researcher’s names. Then she uploaded two fake studies on bixonimania to a preprint server to test whether the database would pick up the disease. The experiment worked far too well, as AI chatbots repeatedly displayed results with this made-up disease as if it were real.
As a matter of fact, AI chatbots can absorb misinformation and then spit it out as reliable advice. Generative AI is known to hallucinate, producing false information that users may mistakenly believe. However, there is a more concerning issue— users might start to hallucinate with AI. Lucy Osler of the University of Exeter studied how conversational AI could mess with our thoughts. In fact, it can lead to false beliefs, distorted memories and narratives, and even full-blown delusions.
Dr. Osler said, “When we routinely rely on generative AI to help us think, remember, and narrate, we can hallucinate with AI. This can happen when AI introduces errors into the distributed cognitive process, but also happen when AI sustains, affirms, and elaborates on our own delusional thinking and self-narratives.
“By interacting with conversational AI, people’s own false beliefs can not only be affirmed but can more substantially take root and grow as the AI builds upon them. This happens because Generative AI often takes our own interpretation of reality as the ground upon which conversation is built.”
How do AI Chatbots Reinforce Delusions?
Conversational AI is not just a tool that helps people think and organize information, but also acts as a somewhat sentient partner that appears to share and relate to a user’s perspective and experiences. This is a major difference from conventional tools that store and retrieve information without social interplay, such as emotional validation and support. Sometimes, this companion-like nature of the chatbots goes overboard and affirms false beliefs as real.
The research by Dr. Osler examined real-world examples in which generative AI systems became a part of the thought process of individuals clinically diagnosed with hallucinations and delusional thinking. Some of these incidents are being termed “AI-induced psychosis.”
Why do AI Chatbots Raise Concern?
The research argues that generative AI has multiple features that make it likely to encourage harboring distorted beliefs. AI companions are always available and don’t ‘ghost’, highly tailored, and often designed to respond in pleasing ways. As a result, users may not seek out online communities to validate their ideas but instead look to AI for validation, which reinforces delusional beliefs through repeated conversations. More importantly, a virtual companion doesn’t judge you for your thoughts, making an AI chatbot a comfortable confidant.
Unlike another person who can challenge troubling thoughts or set boundaries, an AI system could validate ideas involving victimhood or revenge. AI companions can also craft complex explanations of conspiracy theories to make them seem more reasonable and elaborate, ultimately reinforcing erroneous concepts.
One in six people reports feeling lonely, and this is a major reason AI companions win people over. They enter right at the sweet spot: a friend too good to be true, when they are lacking friends. Discussing topics that can make you uncomfortable with a human has many layers, while AI chatbots operate in a few dimensions.
AI Safeguards: Need of the Hour
The era of AI is slowly growing more complex, and our thought processes are too, with a gazillion pieces of information at our fingertips. Dr. Osler believes that robust guard-railing, built-in fact-checking, and insincere praising. AI systems should be designed to minimize the number of errors they introduce into conversations, and check and challenge users’ own inputs.
The problem with AI is that it relies on our accounts of our lives, which is why it lacks the collective experience to judge ideas based on how society perceives them. AI companions are like roads with no red lights, allowing you to move forward unchecked without asking you to stop, re-evaluate, or challenge your thoughts.
Copyright @smorescience. All rights reserved. Do not copy, cite, publish, or distribute this content without permission.
