Olivia Tai's "What We Tell AI" project reveals a growing cultural phenomenon: widespread, often hidden reliance on artificial intelligence for deeply personal decisions. Over the past year, Tai collected more than 350 handwritten confessions from anonymous participants describing how they deploy AI tools in dating, career advancement, creative work, and everyday problem-solving.
The confessions expose a tension between public skepticism about AI and private adoption. People openly embrace generative AI models like ChatGPT and Claude for tasks they rarely discuss openly. Dating profile optimization, job application letter writing, and therapeutic venting appear frequently in Tai's collection. Some confessions reveal users treating AI as a substitute for human judgment in moments requiring discretion or emotional support.
This gap between stated concerns and actual behavior matters for technology companies, policymakers, and investors tracking AI adoption rates. Consumer surveys consistently show Americans express wariness about artificial intelligence. Yet usage data from OpenAI, Google, Anthropic, and Microsoft reveals explosive growth in daily active users accessing these tools. Tai's project quantifies the psychological comfort users feel when the relationship remains anonymous and consequence-free.
The confessions also indicate shifting norms around authenticity and labor. Workers admit using AI to enhance or entirely draft professional communications, raising questions about workplace ethics and competitive advantage. Dating app users confess to AI-generated conversation starters and profile descriptions, blurring the line between authentic self-presentation and algorithmic performance. These behaviors cluster around activities where users perceive high stakes but limited human feedback.
Tai's methodology matters too. The handwritten format creates psychological distance from the digital tools being confessed about, while anonymity removes social friction from admission. This design choice mirrors confessional art projects and church confession booths, suggesting AI has become something people seek absolution for rather than celebrate openly.
The broader implication: AI adoption is already mainstream, regardless of public sentiment. Users compartmentalize their AI dependence, maintaining a public persona skeptical of automation while privately outsourcing decisions to machine learning models. This separation between stated and revealed preferences complicates market research and regulatory frameworks built on survey responses.
For investors and tech leaders, the project signals that concerns about AI adoption faltering due to user backlash may be overblown. Actual usage suggests acceptance happens faster when users perceive personal benefit with low social cost. Privacy and anonymity lower the barrier to AI integration. Companies that preserve user confidentiality while expanding AI applications may capture value faster than competitors emphasizing transparency.
The confessions also hint at emerging psychological effects of ubiquitous AI. Users who outsource decision-making to algorithms may experience decision fatigue differently than previous generations. They may rely more heavily on AI feedback loops, creating feedback systems that reinforce existing preferences rather than challenging them.
As AI tools proliferate across consumer and enterprise applications, Tai's documented shift from public skepticism to private reliance becomes a baseline assumption rather than an anomaly. The year of confessions suggests the real conversation about AI adoption already happened quietly in chat interfaces and document editors, not in town halls or policy debates.