AI Adoption Is Up. Enthusiasm Is not. What that means for your organization.
Last week, Pew released data that showed that nearly half of Americans now use AI chatbots. That’s a sharp uptick from just 33% two years ago.
But despite increased usage, people’s sentiment about AI remains generally negative. 63% of respondents say it's moving too fast, 59% don't trust the companies building it, and 40% think it's bad for society.
Pew does general-public surveys, not workplace studies. But your workplace is probably full of people who feel this way.
An AI transformation strategy that doesn’t account for how people feel is destined to fail. Based on the data and our work helping organizations adopt AI, here are three reasons why sentiment doesn’t always track adoption, and what they mean for your organization.
1. Adoption is often done to people, not chosen by them.
A lot of adoption over the last two years has been driven by AI features coming online in products people already use. (I’m writing this post in Google Docs; I can feel Gemini lurking.)
AI search summaries are a good example. In Pew’s survey, their adoption was even higher than AI chatbots at 60%, which makes sense: Google simply turned that feature on one day.
The problem with adoption by default is that people never actually decided to use these tools, and people rarely warm to something they didn't choose. At best, they tolerate it. At worst, they resent it.
What this means for your organization: lead with pull, not push. Find a real problem to solve, so adoption is something people want rather than something imposed.
2. Vague risks create real anxiety.
Most Pew respondents think AI is moving too fast (63%), driven by companies they don’t trust (59%).
AI doesn't feel like a normal technology. It feels like a giant, unstoppable force. When something feels that big and that uncontrollable, the (very reasonable) response is fear.
In reality, most of the risks organizations face from AI are known and manageable. Hallucinations, environmental costs, and bias are all concrete problems with concrete mitigations.
What this means for your organization: don't pretend AI is risk-free, and don't ask your team to simply “embrace the future.” Name the risks and explain the plan to manage them.
3. The personal case for AI is concrete. The collective one is not.
Among people who use AI, far more say it helps them than hurts them. This is true for their productivity (30% vs. 5%), how informed they are (28% vs. 5%), and even their creativity (21% vs. 11%). Yet only 16% of Americans think AI will be good for society.
That’s because the personal benefits are concrete and immediate, but the collective benefits are vague.
When you send an email a bit faster, or automatically generate meeting notes, the gains are small, but they’re concrete. Compare that to Sam Altman gesturing at an “age of abundance,” or other executives promising AI will "unlock human creativity." It's hard to feel good about a benefit described that vaguely.
What this means for your organization: AI is a way to do work faster, but pitching that way won’t win your team over. Instead, ask what your organization can do with AI that it couldn't before, and use that answer to build your strategy and your internal story.
***
Earlier this year, some companies installed leaderboards that ranked employees by their AI usage. “Tokenmaxxing” became fashionable. The assumption was that more AI use meant more productivity.
Goodhart's Law says that "when a measure becomes a target, it ceases to be a good measure." For all those tokenmaxxing companies, usage didn't reliably predict productivity. It did reliably predict large invoices from OpenAI and Anthropic.
If your AI strategy cares only about adoption, you're only solving half the problem. The other half is helping people understand how to navigate the change and what success looks like.
Last week, Pew released data that showed that nearly half of Americans now use AI chatbots. That’s a sharp uptick from just 33% two years ago.
But despite increased usage, people’s sentiment about AI remains generally negative. 63% of respondents say it's moving too fast, 59% don't trust the companies building it, and 40% think it's bad for society.
Pew does general-public surveys, not workplace studies. But your workplace is probably full of people who feel this way.
An AI transformation strategy that doesn’t account for how people feel is destined to fail. Based on the data and our work helping organizations adopt AI, here are three reasons why sentiment doesn’t always track adoption, and what they mean for your organization.
1. Adoption is often done to people, not chosen by them.
A lot of adoption over the last two years has been driven by AI features coming online in products people already use. (I’m writing this post in Google Docs; I can feel Gemini lurking.)
AI search summaries are a good example. In Pew’s survey, their adoption was even higher than AI chatbots at 60%, which makes sense: Google simply turned that feature on one day.
The problem with adoption by default is that people never actually decided to use these tools, and people rarely warm to something they didn't choose. At best, they tolerate it. At worst, they resent it.
What this means for your organization: lead with pull, not push. Find a real problem to solve, so adoption is something people want rather than something imposed.
2. Vague risks create real anxiety.
Most Pew respondents think AI is moving too fast (63%), driven by companies they don’t trust (59%).
AI doesn't feel like a normal technology. It feels like a giant, unstoppable force. When something feels that big and that uncontrollable, the (very reasonable) response is fear.
In reality, most of the risks organizations face from AI are known and manageable. Hallucinations, environmental costs, and bias are all concrete problems with concrete mitigations.
What this means for your organization: don't pretend AI is risk-free, and don't ask your team to simply “embrace the future.” Name the risks and explain the plan to manage them.
3. The personal case for AI is concrete. The collective one is not.
Among people who use AI, far more say it helps them than hurts them. This is true for their productivity (30% vs. 5%), how informed they are (28% vs. 5%), and even their creativity (21% vs. 11%). Yet only 16% of Americans think AI will be good for society.
That’s because the personal benefits are concrete and immediate, but the collective benefits are vague.
When you send an email a bit faster, or automatically generate meeting notes, the gains are small, but they’re concrete. Compare that to Sam Altman gesturing at an “age of abundance,” or other executives promising AI will "unlock human creativity." It's hard to feel good about a benefit described that vaguely.
What this means for your organization: AI is a way to do work faster, but pitching that way won’t win your team over. Instead, ask what your organization can do with AI that it couldn't before, and use that answer to build your strategy and your internal story.
***
Earlier this year, some companies installed leaderboards that ranked employees by their AI usage. “Tokenmaxxing” became fashionable. The assumption was that more AI use meant more productivity.
Goodhart's Law says that "when a measure becomes a target, it ceases to be a good measure." For all those tokenmaxxing companies, usage didn't reliably predict productivity. It did reliably predict large invoices from OpenAI and Anthropic.
If your AI strategy cares only about adoption, you're only solving half the problem. The other half is helping people understand how to navigate the change and what success looks like.