hough Silicon Valley might have a rosy picture of what AI can do for people, the broader narrative around the technology is murky. So Anthropic sought clarity.

On Wednesday, Anthropic published the results of a December survey of more than 80,000 Claude users from 159 countries and 70 languages, who conversed with an AI interviewer built by Anthropic. The survey, which The Deep View’s Sabrina Ortiz took part in during her time at ZDNET, sought to illuminate the desires, fears and reality of AI to better define the “abstract projections of its risks,” Anthropic said.

Broadly, users crave a tool that can manage their work, their life and their time. While many clocked major time savings, thousands reported that AI isn’t making the grade just yet.

Here are the highlights:

• What users want: Almost 19% of users reported wanting a tool to help them professionally by handing off routine tasks so that they could focus on more complex work. However, large percentages of users also reported wanting tools to help with life management and personal transformation, with those categories coming in at 13% each.
• What users fear: The biggest concern was reliability, with more than a quarter (27%) of users surveyed fearing domino effects regarding inaccuracy and hallucinations. But beyond this, users largely feared AI doing more than it should, with respondents reporting concerns of job displacement, autonomy and cognitive atrophy.
• The reality: The biggest benefit that users have realized by far has been productivity, with 32% reporting that the tech has dramatically sped up work and tedious tasks. However, 19% reported that the tech hasn’t delivered on its promises and isn’t yet capable of doing what they envisioned.

The goal of this report was to help Anthropic identify what it’s doing well and what it could be doing better. The throughline, the company said, is “that AI helps them live better, not simply work faster.”

Anthropic’s study begs the question of how people really feel about AI. Surveys from organizations like Pew Research tell a different story: Americans are more concerned than excited about AI, believing it will worsen our ability to think for ourselves and harm our ability to form meaningful relationships.

Our Deeper View

Though Anthropic’s survey is one of the largest research studies on AI ever conducted, it may have a blind spot: It only studied Anthropic users. Because Anthropic’s report seeks to inform Anthropic about its own shortcomings, we should be wary about taking this data at face value. The narrative around AI among the broader public may not be nearly as enthusiastic, with many distrusting and fearing the technology. However, for these companies to succeed in the long term, especially among consumers, they know they need to shift that perception.