OpenAI's GPT-4o Makes AI Clones of Real People With Surprising Ease
Artificial Intelligence (AI) has made remarkable progress in mirroring human conversational abilities. Recent research has unveiled a more astonishing capability of AI - the ability to replicate specific individuals' personalities.
Cloning Personalities
Humans are complex beings, shaped by a combination of nature and nurture over years of unique experiences. However, a study conducted by researchers at Stanford University suggests otherwise. According to the study, it only takes a two-hour interview for an AI model to predict people's responses to various tests and questionnaires with an impressive 85% accuracy.
While the concept of cloning personalities may seem eerie, the researchers believe it could serve as a valuable tool for social scientists and policymakers. By simulating responses to different policy choices, these AI models offer a new way to understand human behavior.
Creating AI Replicas
The study utilized a voice-enabled version of OpenAI's GPT-4o to interview over 1,000 individuals across the United States. The AI model asked scripted questions and engaged in follow-up inquiries based on the participants' responses. Subsequently, AI agents were trained to mimic the participants' answers by analyzing the interview transcripts.
Comparative tests between the AI agents and human participants, including social attitude surveys, personality assessments, economic decision-making games, and social experiments, revealed an impressive accuracy rate of 85% after accounting for human response variations.
Implications and Concerns
The potential applications of creating AI replicas of individuals are vast, ranging from enhancing policymaking processes to conducting cost-effective focus groups. However, there are concerns about the misuse of this technology for deceptive purposes, as demonstrated by previous incidents of deepfake scams.
Despite the ethical dilemmas, the research indicates that machines capable of realistically emulating humans in diverse scenarios are on the horizon.
