For consumers, the AI race is starting to feel less like a competition over who has the smartest model and more like a hunt for where the intelligence shows up. PYMNTS research finds a clear split between “AI you go to,” such as ChatGPT and other dedicated platforms, and “AI that comes to you,” which lies inside apps, phones and merchant websites.
That distinction is becoming important as consumers appear to be finding more value in purpose-built AI tools, especially for health, learning and everyday planning. At the same time, embedded AI is losing ground across many routine tasks. The data also shows how much habit and an AI tool’s location shape whether consumers use it. For example, Android users are more likely to use Google Gemini because it’s built into their mobile experience. iOS users tend to lean more heavily into mainstream AI use and are more likely to choose ChatGPT. The result is a consumer AI market being shaped as much by trust, familiarity and default locations as by the technology itself.
The main finding from this latest PYMNTS Intelligence Consumer AI Benchmark Report is that consumers still use AI heavily for writing, search and research. But the next growth wave is forming around repeatable daily chores, including meals, shopping, reminders, budgeting, household logistics and personal planning. The frontier is shifting from asking AI for information to letting AI help manage the routines that fill the day.
AI in Everyday Life
Consumer AI use is moving into daily routine management.
Across the 54 personal activities tracked in March 2026, writing remains the most common reason consumers turn to AI. Among consumers who used AI for at least one personal task, 31% used it to edit or reword personal writing, while 30.1% used it to draft texts or emails. Product discovery is close behind, with 29.6% using AI to find product links to purchase, followed by 27.4% who used it to look up symptoms or conditions and 26.6% who used it to plan meals or grocery lists.
Those top-line rankings show where AI is already a part of everyday life. The movement underneath them shows where it’s becoming more useful. From the October-November 2025 average to March 2026, the fastest-growing AI tasks were finding discount codes or deals, up 2.9 percentage points; managing household logistics, up 2.8 points; planning meals or grocery lists, up 2.7 points; decorating or home improvement tips, up 2.5 points; and finding products connected to celebrities or influencers, up 2.4 points.
As AI usage becomes more practical, travel-related tasks are moving in the opposite direction. Travel safety or visa information fell 2.2 points, finding hotels, restaurants or attractions slipped 1.7 points and translating customs or phrases slid 1.3 points. It’s possible that affordability, timing or changing travel demand could be influencing those declines, but the broader pattern still shows household and shopping tasks gaining, while less frequent travel tasks pull back.
For banks, merchants and payments providers, that shift is worth tracking because many of the fastest-growing activities sit close to spending decisions. Discount searches, grocery planning, bill reminders, household budgets and product discovery all touch commerce or financial management. In those use cases, AI is starting to shape how they plan, compare, buy and remember.
The most helpful AI tools
However, the report includes an important warning about embedded AI. While more consumers are encountering AI inside apps and websites, fewer are naming those tools as the most helpful.
AI embedded in third-party or merchant apps and websites lost ground as the “most helpful” tool in six of eight task groups, with the steepest declines in learning and self-improvement (down 12.1 percentage points), health and wellness (down 7.1 points) and everyday planning (down 5.5 points). Dedicated AI platforms gained in health and wellness (up 6.2 points), everyday planning (up 3.9 points), shopping and purchasing (up 3.5 points) and finance and banking (up 3.1 points).
That finding challenges a common assumption about embedded AI. A merchant app may know its own inventory best, but consumers often start with discovery rather than a specific merchant. The report shows that consumers may be going to dedicated AI platforms because they want comparison, context and breadth before they narrow the choice to a particular merchant.
Android or iOS?
Defaults shape consumer AI choice, but engagement skews toward Apple.
Operating systems are becoming one of the most important distribution paths for consumer AI. Among dedicated AI platform users, Android users are far more likely than iOS users to use Google Gemini. The gap is 24 percentage points, with 73% of Android users reporting Gemini use, compared with 49% of iOS users. ChatGPT shows the reverse pattern, with 88% of iOS dedicated platform users reporting ChatGPT use, compared with 78% of Android users.
The Gemini-Android gap stands out because other AI platforms show much smaller operating system differences. Microsoft Copilot is used by 40% of Android users and 33% of iOS users. Meta AI is almost even, at 33% for Android and 32% for iOS. Claude, Perplexity and Grok also show narrower spreads. Against that backdrop, Gemini’s advantage among Android users looks less like a general model preference and more like the power of default placement.
Built into Google’s mobile ecosystem, Gemini sits near Gmail, Maps, Calendar, Wallet and other daily tools. That proximity can create a cycle of familiarity. Consumers see the tool, try it, become more comfortable with it and may eventually trust it with more tasks. This can be framed as an ecosystem advantage rather than a pure technology advantage.
Consumer choice of operating system impacts AI activity personas.
Yet the operating system story doesn’t end with Android’s distribution edge. iOS users are more likely to be active AI users than Android users. PYMNTS finds that 52% of iOS users are power or mainstream AI users, compared with 40% of Android users. Android users are also more likely to be non-users, at 49%, compared with 37% of iOS users.
The desktop pattern points in the same direction. Sixty percent of macOS users are active AI users, compared with 46% of Windows users. Mac users also have a higher share of power users, at 21%, compared with 12% for Windows.
For Google and Apple, the data creates a split strategic picture. Google has a stronger built-in surface for proactive AI through Android, but Android also contains a larger pocket of non-users to convert. Apple has a more AI-engaged user base, but its native proactive AI experience has not matched Google’s reach. That gap helps explain why iOS users lean toward ChatGPT. They may be more active with AI, but they often choose a dedicated AI brand rather than relying on the phone’s native assistant.
Anticipatory and On-Demand AI
Consumers still want control, except for the heaviest AI users.
When consumers are asked how they want AI to interact with them, the largest single group still prefers to be in charge. PYMNTS finds that 41% want “AI you go to,” meaning they prefer to activate AI themselves. Another 36% prefer “AI that comes to you,” meaning AI that surfaces information automatically, while 23% have no preference.
That topline result can make the market look cautious. A closer look shows a more segmented market that depends on how frequently a consumer uses AI.
Among power users, 55% prefer anticipatory AI. Among light users, only 18% prefer that model, while 50% want to ask AI themselves. Mainstream users sit closer to the overall market, with 36% preferring anticipatory AI and 42% preferring on-demand AI.
Age sharpens the divide. Gen Z consumers are the most open to anticipatory AI, with 48% preferring AI that comes to them. Millennials follow at 42%. The share falls to 36% among bridge millennials, 28% among Gen X and 21% among boomers and seniors. Among boomers, 61% prefer to initiate AI themselves, the strongest single preference in the data.
Operating systems add another wrinkle. iOS users are 7 percentage points more likely than Android users to want AI that comes to them, 36% versus 29%. On computers, macOS users are also more likely than Windows users to want anticipatory AI, 36% versus 31%.
This finding reframes Gemini’s Android advantage. Gemini’s lead among Android users appears tied to distribution, while latent demand for proactive AI may be stronger among Apple users. In other words, Google has more built-in delivery. Apple has a larger audience that is already ready for this mode.
Permission, Privacy and Usefulness
The divide between on-demand and anticipatory AI becomes clearest when consumers explain why they prefer one model over the other.
Among consumers who want AI to surface information automatically, the top reason is practical: 51.5% say it would save them more time. The next reasons, however, point to a broader shift in expectations.
Consumers say anticipatory AI feels more natural (41.5%) and more like a real assistant (40%). Another 32.3% say they trust AI to highlight what’s relevant, while 30.8% say it would handle the mental load of remembering.
Those responses describe AI as more of a helper than a search box. A proactive assistant could surface a flight detail while a consumer is texting about a trip, send bill reminders, suggest a grocery list, point out a deal or pull up information before the user has to ask. The value is not only that the task gets done faster. It’s that the consumer doesn’t have to keep track of every step.
Among consumers who prefer on-demand AI, the resistance is not about whether AI is smart enough. Only 20% cite a lack of confidence that it would get things right. The dominant concerns are control and privacy. Nearly three-quarters (74.1%) say they want to retain control over when AI is involved. More than half (56.1%) say they don’t want AI monitoring their activity or data. Another 41.8% say they don’t want AI interrupting what they’re doing, and 39.3% say unsolicited suggestions are annoying.
Experience changes the picture. Consumers already using AI embedded in apps or websites are the strongest proponents of anticipatory AI, with 51% preferring that model. Consumers who have fully or mostly replaced older methods with AI are also more likely to prefer anticipatory AI, at 49%. By contrast, 27% of traditional search users and 25% of consumers who use AI only as a light complement to older methods prefer AI that comes to them.
Key differences
The generational data shows that consumers can want the same interaction model for different reasons. Among Gen Z consumers who prefer anticipatory AI, 20% say handling the mental load is the most important reason, tied with saving time. For boomers who prefer anticipatory AI, saving time leads at 31%, followed by the idea that AI feels like a real assistant at 29%. In the internal discussion, the team noted that for older consumers, this may resemble reminders and practical support: a tool that helps them remember appointments, obligations or next steps.
On the on-demand side, the generations are more aligned. Staying in control is the leading reason across Gen Z, millennials, Gen X and boomers. Concern about background monitoring is also consistently high across age groups. That makes control the universal design requirement, even for products aimed at younger users.
Operating system differences show where trust may live. Among consumers who prefer on-demand AI, iOS users are more likely than Android users to cite concerns about AI monitoring or data, at 61% versus 52%. Among those who prefer anticipatory AI, Android users are more likely to say proactive AI feels natural, 47% versus 39%. When asked whom they would trust to provide an AI assistant, 31% of Android users and 23% of iOS users say they would never trust anyone. AI-first brands, smartphone manufacturers, big tech platforms, digital wallets, and banks or card providers each attract smaller shares.
Conclusion
Consumer AI is entering a more practical phase as the initial novelty of general-purpose tools gives way to everyday use. Adoption has broadened but not surged, and the early-adopter capture appears largely complete. The more useful story is how consumers are deciding where AI belongs in their routines.
Dedicated AI platforms are gaining credibility for tasks that require breadth, judgment or personal context. Embedded AI has reach, but its position inside an app or website is not enough. Consumers appear willing to use AI in more parts of daily life, but they still expect the tool to justify the interruption, the data access and the trust.
The next competitive advantage will likely belong to providers that offer both modes: an AI that consumers can ask directly and an AI that can step forward when permission has been granted. Power users and younger consumers are already leaning toward the second model. Most consumers still prefer the first. The winners will let consumers move between the two without feeling watched, interrupted or pushed.
Read More
PYMNTS Intelligence is the leading provider of information on the trends driving consumer use of AI in their daily lives. To stay up to date, subscribe to our newsletters and read our in-depth reports.
Methodology
“The New AI Handshake: Data Shows When Consumers Want Help and When They Want Control,” the seventh installment of the PYMNTS Intelligence Agentic AI Series, is based on a March 2026 survey of 2,111 U.S. adults. The report examines consumer use of AI for personal tasks across nine activity categories and 54 distinct activities. It also explores consumer preferences for proactive versus on-demand AI interaction models. The activity persona framework classifies consumers into power users, mainstream users, light users or non-users based on the breadth and complexity of AI tasks they perform. Samples were balanced to match the U.S. adult population by age, gender, education and income.
