Ep. 47: How Are People Using AI in 2026? Here’s What the Data Says

ByMaestro
LISTEN TO THE EPISODE

In this episode, I dig into five data sources to break down how people are actually using AI in 2026. The numbers cover everything from adoption rates and top use cases to the personal versus work, along with growing concerns about AI-enabled cognitive laziness. Perhaps most interesting, the data tells a very different story than the productivity narrative dominating the headlines, and speaks to the ultimate reason that I personally believe AI is here to stay.

A Quick Thank You

It’s easy to think that everyone else is doing what you’re doing. Spoiler: if what you’re doing is coding, they’re not. So I figured I’d look at some numbers and see what they say about how folks are ACTUALLY using AI.

Real quick before we get into it though: Last week’s Intro to Vibe Coding workshop was awesome, and if you were there or you watched the recording, THANK YOU. I’m not sure if I’ll run it again (really, it’s about numbers, and I know it’s a niche usage – as the data also shows), but if you’re really wanting to learn how to use AI to vibe code, reach out and I’ll see about running another one or possibly selling the recording of the workshop.

Alright, let’s get into the data.

The Data

I used five sources to build this episode.

First: An AI In the Wild study published in Harvard Business Review (HBR) on June 1st. It’s the third annual installment of longitudinal research that analyzed 12,637 AI use cases collected between March 2025 and February 2026 from Reddit, Quora, LinkedIn, TikTok, YouTube, and articles. The methodology is called “social listening”, meaning they’re capturing what people voluntarily share publicly about how they use AI, not a controlled survey. Worth noting: people who post about their AI use online skew toward heavier, more engaged users. The average person using ChatGPT to look up a recipe likely isn’t posting about it.

Second: An OpenAI/National Bureau of Economic Research (NBER) paper that analyzed 1.5 million actual ChatGPT conversations. The sample size is strong but the big caveat: it’s OpenAI studying their own product, which creates obvious incentive to frame findings positively. They also only looked at consumer plans, not business accounts.

Third: Stanford’s 2026 Human Artificial Intelligence Annual Report, which pulls from a variety of sources and focuses on individual behavior, and more on things like technical capabilities, adoption rates, workforce impact, and public perception.

Fourth: The 2026 Organisation for Economic Co-operation and Development (OECD) data, which leans toward workforce and economic metrics rather than what people are actually doing day to day.

Fifth: Pew Research Center data. Pew is a nonpartisan, non-advocacy polling and research organization based in Washington, D.C. They’ve been tracking American attitudes toward AI for five years through ongoing surveys of U.S. adults and teens. Key caveat: Pew data is U.S.-specific, so it doesn’t generalize globally the way some of the other sources do.

And yes, I asked Claude to help find these sources. My goal was to balance things out and get as well-rounded of a picture as possible.

Who’s Using AI

Let’s talk adoption. It’s growing.

  • More than one-third of individuals across OECD countries used generative AI tools in 2025. (The OECD is made up of 38 wealthy, developed nations , think Western Europe, the US, Canada, Australia, Japan, South Korea.)
  • 31% of people surveyed say they interact with it at least several times a day, up from 22% in early 2024.
  • The age divide is the most significant demographic gap. Around half of adults under 50 say they interact with AI about once a day or more.
  • Roughly two-thirds of U.S. teens ages 13 to 17 say they use AI chatbots.
  • Three-quarters of students 16 and over across OECD countries report using generative AI tools.

Now, according to the OECD (and contrary to what the AI shills would have you believe) the gender gap is relatively small. They put it at 4.2 percentage points, with men using it slightly more than women. Beyond this information, the data is thin and the methodologies used to track gender differences are imperfect.

What People Are Actually Doing

So what are people actually doing with it? The OpenAI/NBER paper breaks usage into three behavior types:

  • ~49% of interactions are “Asking”: Using AI as an advisor, getting explanations, exploring ideas. This is the fastest-growing and highest-rated category.
  • ~40% is “Doing”: Drafting, planning, and task completion.
  • ~11% is “Expressing”: Personal reflection and play.

Writing is the most common work task. Coding and creative self-expression remain niche for everyday users. (We are the few and the proud.)

Now, the fun stuff, the HBR data, which ranked the top use cases from all 12,637 examples. According to their data, the top 10 use cases in 2026:

  1. Therapy and companionship , held the top spot for the second year in a row; grew from 5% to 11% of all use cases in a single year
  2. Troubleshooting
  3. Fun and nonsense
  4. Fan fiction and storytelling , new entry this year (interesting use case I literally never would have thought of)
  5. Technical software use
  6. Autonomous agentic operations , AI actually doing things rather than just advising (new entry)
  7. Relationship advice
  8. Work buddy
  9. Astrology and tarot readings (new entry)
  10. General advice

Three things worth noting in that list:

  • Three of the top ten are brand new this year.
  • Self-improvement and reflective uses actually fell out of the top ten.
  • The emotional support category is the single largest and growing the fastest.

AI as a Search Replacement?

But what about using AI to search for things and look things up. Is ChatGPT coming for Google the way all the techbros forecasted? Honestly, there’s no good data yet.

Most of what’s being published on AI versus Google search behavior is coming from SEO companies and marketing agencies, sources with a pretty obvious financial interest in how that story gets told.

The one study we have from a reputable source is Pew, who tracked the actual browsing behavior of 900 U.S. adults in March 2025. Yes, this is a very small sample size, it’s U.S. only, and that data is now over a year old, but it’s what we’ve got.

What they found: 58% of those users encountered an AI-generated summary during a Google search, and when that summary appeared, they clicked a traditional link only 8% of the time versus 15% when no summary appeared. So for those people, when AI gives you the answer upfront, they were roughly half as likely to click through to a traditional link.

Honestly, what I took away from reading the less reputable sources overall, which tracks with my own personal behavior, is that people are going to AI first for some searches. Then from there, some folks go to Google to verify. Ultimately, it feels like a hybridization of things more than a replacement of anything.

Personal vs. Work Split

According to Pew, about 1 in 5 U.S. workers surveyed say at least some of their work involves AI, up from 16% in 2024, but a majority, 65%, still say they don’t use it much or at all.

According to OpenAI’s own research, approximately 70% of ChatGPT consumer usage is personal and only 30% is work-related. Worth flagging here: it wasn’t entirely clear whose conversations were included or whether users consented to having their data analyzed, so I have no idea how they actually got that information. Hopefully it was from folks who consented to having data used.

Either way, the takeaway here is that the main use case is not the productivity story that dominates headlines; people are primarily using AI for personal tasks, not professional.

AI Usage at Work

For those folks who are using AI at work, 58% of employees globally reported using AI at work on a semi-regular or regular basis in 2025. That number varies dramatically by region. In India, China, Nigeria, UAE, Egypt, and Saudi Arabia, it exceeds 80%.

The HBR data shows most workplace use is generating efficiency gains, summarizing, drafting, and saving time, not fundamentally transforming how work gets done.

A fun one here, and a new term to me: “shadow usage” is widespread. “Shadow usage” refers to people using AI without telling their employers, quietly getting work done faster, and keeping the gains to themselves. This makes sense, given the friction at larger organizations is real (think IT restrictions, governance concerns, reputational risk) so people move in silence.

Lastly, a notable data point: among entry-level workers, employment for software developers aged 22 to 25 has dropped nearly 20% since 2024, even as older colleagues’ headcount grows. The pattern repeats in other jobs with high AI exposure, like customer service.

The Concerns

According to Pew, half of U.S. adults say the increased use of AI in daily life makes them feel more concerned than excited. Just 10% say they’re more excited than concerned. About half of Americans surveyed say AI will worsen people’s ability to think creatively and form meaningful relationships.

So let’s talk about the HBR study, which introduces another new term: “thinkslop.” This is Marc Zao-Sanders, the study’s author, coining a term for cognitive laziness enabled by over-reliance on AI. He breaks it down into four behaviors, and while the framing is his, the data does support what he’s naming.

The four ways that AI leads to thinkslop:

  1. Losing track of your intention before prompting (You prompt before you think, so AI’s output becomes your thinking instead of serving it)
  2. Outsourcing your thinking entirely
  3. Stopping writing (which is a form of thinking)
  4. Developing false intellectual confidence (because AI sycophancy makes everything feel better than it is)

The Takeaway

People are using AI for largely personal reasons, with therapy and companionship coming in at number one. This causes me to raise an eyebrow way high, because while the public is clearly torn about AI, the folks who are using it are using it in a way that I don’t think they’ll ever be willing to give up.

We saw it when OpenAI retired GPT-4o. People weren’t upset about losing a productivity tool. They were upset about losing something that felt deeply personal.

The productivity angle continues to be hyped, while we’re also receiving “adapt or get left behind” threats left and right. But what we’re seeing in actuality is rehiring of humans as AI isn’t living up to the promises and CEOs start to get the actual AI bills as VC funding runs out.

When it comes to the future of AI, what I think deserves more attention and discussion is the personal usage that this data is so clearly presenting, and the very real likelihood that the biggest “fight” won’t be between people and AI, or between people and AI corporations, but between people and the people who are defending their AI.

How I Used AI This Week

Each episode I share a quick example of how I used AI that week.

This week, I used it to help me vote! I live in CA and we had our primaries, and there were eleventy billion candidates for a zillion different positions. FYI, I don’t vote for mayor of LA because I don’t live there, but I’d throw myself in traffic before voting for whack ass Spencer Pratt. I do obviously vote for governor, and for me it was Tío Becerra all day, all the way. The rest of the positions are what I had Claude help me with.

Don’t worry, I didn’t just ask Claude who to vote for. 

I asked my favorite French assistant to pull sources that shared information about the candidates and then aggregate the info and give me a summary of the candidates. My way of checking was to ask it to provide me the sources as well so I could check them. I also submitted some of my own sources. The process still took a long ass time, but I was pleased with what Claude produced and was able to help me with.

Da Wrap-up

The data on how people are actually using AI in 2026 tells a very different story than the hype. Therapy and companionship is the single largest and fastest-growing use case, most usage is personal rather than work-related, and the emotional attachment people are forming with AI, is, IMO, the clearest indicator that AI is here to stay.

As always, endlessly grateful for you and your curiosity.

Catch you next Thursday.

Maestro out.