We Don't Trust AI. We Just Can't Stop Using It.

There is a striking contradiction at the heart of how people relate to artificial intelligence today: widespread skepticism coexisting with near-universal daily use. New research makes this tension impossible to ignore.

Three studies released in recent months - from OpenAI, Anthropic, and Ipsos - paint a more nuanced and surprising portrait of how we actually use these tools. For legal professionals, the findings carry real strategic implications.

The Numbers That Surprised Everyone

The most striking finding: 73% of ChatGPT conversations have nothing to do with work. That figure has jumped sharply compared to last year. What was marketed as a professional productivity tool has quietly become something more personal - a writing assistant, a thinking partner, a faster alternative to a Google search.

Anthropic, by contrast, reports growth in professional and work-related use of its models. This is not necessarily a contradiction. It may simply reflect that different tools attract different use cases - and that the market is segmenting in ways that matter for anyone advising organizations on AI adoption.

The Three Dominant Use Cases

Across all the research, the top three uses are decidedly everyday: writing assistance (emails, messages, anything that needs a polish), practical guidance (everyday advice, quick tips, working through small problems), and fast information retrieval (a more conversational replacement for a search engine).

None of these are the high-stakes, transformative applications that tend to dominate the headlines. That gap between expectation and reality is worth examining.

Who Is Actually Using These Tools

The demographic shift is significant. Within a single year, the share of women among AI users jumped from 37% to 52%. Women show a stronger preference for writing and practical guidance; men lean toward technical and multimedia applications. Adoption still skews toward younger, highly educated men overall - but the gap is closing fast.

The Trust Gap: Saying One Thing, Doing Another

Only half of the public trusts AI providers to develop these systems responsibly. Yet the same people open these tools every day - for their most personal communications, their most private questions. The gap between stated skepticism and actual behavior is not hypocrisy. It is a rational response to tools that are genuinely useful, regardless of whether we fully trust the companies behind them.

For lawyers and in-house counsel, this dynamic should sound familiar. Clients ask whether they can trust AI. The more useful question is how to govern its use.

Enterprise Adoption Is Harder Than It Looks

The research confirms what practitioners already suspect: deploying AI inside an organization is not a switch you flip. It requires substantial investment in employee training, data modernization, and process redesign. The legal profession is no exception.

At a recent AI forum colleagues gathering under the ACC AI Forum - which I have the privilege of leading - colleagues shared candid accounts of the legal technologies already embedded in their departments, alongside honest assessments of the organizational challenges involved. The adoption curve is real. So are the competitive advantages for those who navigate it well.

The Strategic Takeaway for Legal Professionals

The research makes one thing clear: AI is not approaching. It is already here, already embedded in how people work and communicate - including the clients, counterparties, and colleagues you interact with every day.

The choice facing legal professionals is not whether to engage with this technology. It is whether to engage on your own terms, ahead of the curve, or to be pulled along by circumstances once the window for shaping norms has closed. Those who build fluency now - who learn to govern AI use rather than simply react to it - will define the standards of legal practice for the decade ahead.