Americans are testing AI, but they want guardrails

The tech world talks about AI as if humanity has already uploaded its consciousness to the cloud. But outside Silicon Valley and Twitter, what are everyday people actually doing with these tools?

To get a real reality check, The Verge teamed up with Vox Media and research firm The Circus to survey more than 2,000 US adults. The big takeaway: hype is everywhere, but hands-on adoption is still surprisingly narrow.

Everyone has heard the hype, but few actually use it

Only 1 in 3 people have ever touched a dedicated generative AI tool.

OpenAI's ChatGPT is by far the most recognized name on the board, but even then, only 20% of respondents said they have actually used it. Another 37% have heard of it, while 43% have never heard of it at all.

Big tech alternatives drop off fast from there. Microsoft's Bing with ChatGPT and Snapchat's My AI sit at just 12% usage each, while Google's Bard managed only 10%.

For generative image tools, the numbers fall off a cliff. Midjourney has been used by just 7% of people, with 75% never even hearing of it. Stable Diffusion fares worse: only 6% used it, and 77% have no clue what it is.

Unsurprisingly, usage is heavily skewed toward Millennials and Gen Z, with roughly 36 million and 35 million people familiar with the tools respectively. Boomers barely register at 4.8 million.

Yet despite low everyday usage, public expectations are massive. 74% believe AI will have a large or moderate impact on society, handily beating electric vehicles at 69%, VR at 60%, and leaving NFTs in the dust at 34%.

Coding is rare, but style-cloning is everywhere

What are the people who use AI actually doing with it? Mostly searching and fooling around.

68% use AI to answer questions, essentially treating chatbots like upgraded search engines. 54% use them for brainstorming. Professional tasks like coding sit down at just 18%.

Instead, casual creative experimentation dominates: 37% use it for photos, 31% for music, and 31% for video. Crucially, a wide majority said AI helped them create things they could never make on their own, and that the output was better than what they could have done.

That creative power comes with severe hypocrisy around ethics. 70% of respondents agree that artists should be compensated when an AI copies their style. But only 43% support outright bans on derivative creations, and 44% admitted they have personally prompted an AI to mimic an artist, author, or celebrity.

When it comes to rules, people want heavy guardrails. 76% want new government regulations, 78% demand mandatory labels on AI-generated content, and 76% want non-consensual deepfakes outlawed.

Overall, the mood tilts toward anxiety: 68% fear privacy threats, 67% fear corporate or government misuse, and 63% worry about job loss.

Why it matters

This poll exposes the massive gap between tech industry buzz and real-world adoption.

For builders and creators, it is a reminder that standalone AI toys still have a niche audience. Mainstream consumers are not rushing out to register for Discord servers or learn prompt engineering. The winners won't be the standalone model wrappers, but the established software suites that quietly bake AI into existing workflows like Gmail, Docs, and Photoshop.

We think the findings around copyright and ethics highlight a brewing regulatory storm. The public loves the convenience of copying artistic styles on demand, but overwhelmingly demands artist compensation and strict deepfake crackdowns.

AI labs are racing toward superintelligence, but they are operating on borrowed public goodwill. If companies don't build credible attribution, watermarking, and consent systems right now, governments will likely step in with blunt, restrictive laws that slow down the entire sector.