Many AI companies are betting their future on autonomous agents — AI assistants meant to take over mundane tasks like email management, travel bookings and online shopping. Meta’s Muse rose quickly to the top of the Apple App Store thanks to integrations with Facebook and Instagram and a conspicuous mascot; Meta later made Muse available to business customers as well.
Industry moves: Microsoft and OpenAI
Other major players have followed suit. Microsoft upgraded Copilot with an Autopilot feature that embeds more agentic capabilities into workplace tools, and OpenAI announced its own agentic assistant, dots, this week.
The main barrier: user resistance and privacy concerns
Despite these developments, consumer uptake is far from guaranteed. A Pew Research Center survey in February found that 51% of respondents avoid AI chatbots entirely. Among those who don’t use chatbots, 79% cited privacy concerns and 67% said they were unlikely to try a chatbot in the next year.
The deeper obstacle is that agents become genuinely useful only if users grant them significant access to their digital lives — access most people are reluctant to provide. A large international Thales poll conducted this year found only 13% would let an "AI helper" read their email, 11% would let one rebook travel, and just 7% would allow it to move money between bank accounts. YouGov data showed 56% of respondents would not let an agent shop for them at all, and only 10% trust an agent with more than $25 without approval.
Usage skews toward higher-income, tech-adjacent households
Survey work by Menlo Ventures with Morning Consult indicates AI use is strongly correlated with income: households earning over $100,000 are substantially more likely to use AI than those earning under $50,000. Paying customers use AI twice as often daily and are five times more likely to use agents. Menlo summarizes the typical AI spender as "a millennial parent with a post-grad degree working in technology or financial services" who "tends to have more money than time."
Adoption is "widespread but shallow"
The St. Louis Federal Reserve describes AI adoption as "widespread but shallow": at least 20% of workers use AI across more than 80% of occupations, but adoption is concentrated in white‑collar roles. Use rates hover around or above 80% for computer programmers, PR professionals, financial advisers and CEOs, while licensed practical nurses, receptionists and animal caretakers see adoption at or below 10%.
As the St. Louis Fed puts it: "Currently, AI works on screens and cannot drive a truck or draw blood."
Company data reflects the same skew
Anthropic’s Economic Index Survey (which the company notes is not representative) also showed overindexing in computer/math-related jobs and management — professions that make up large shares of respondents but smaller shares of overall U.S. employment.
Societal priorities place AI low on the list
Gallup polling finds only 6% of Americans view using AI tools like ChatGPT as a very important ingredient for a good life. By contrast, 78% prioritize spending time with partners, friends or family; 58% value exercise; 49% value cooking; 45% time in nature; and 42% spirituality.
Bottom line: limited near-term impact outside tech‑adjacent roles
Agentic AI could become transformative over time, but current evidence suggests its near‑term effects will be concentrated among those already familiar with digital productivity concepts like "inbox triage" and overwhelmed by calendar notifications. For broader adoption, agent makers face the twin challenges of convincing users to surrender deeper access to digital lives and addressing persistent privacy and trust concerns.
