We’ve heard everyone from Mark Zuckerberg to Anthony Albanese tell us that AI is going to change the way business operates around the world. New data shows that big business isn’t quite convinced yet.

Big companies are stepping back from AI

Despite the hype, new data from the United States Census Bureau suggests that big businesses are cooling on AI. The Bureau regularly surveys over 1.2 million businesses across the country, asking, among other things, whether they’ve used AI tools (like machine learning, natural language processing, chatbots or voice recognition) in the last fortnight to help produce goods or services.

The most recent survey results in the graph above show a clear trend: while small and medium businesses are still slowly adopting AI, companies with more than 250 employees are doing so at a decreasing rate. In other words, AI use in large firms is not just slowing, it’s declining.

And while the data is US-based, Australia is following a similar trajectory. Surveys by the Future Skills Organisation and the AI Group indicate that local companies, especially larger ones, are treading carefully. They’re testing AI, yes. But widespread implementation? That’s still a way off.

Why workers might not be embracing AI

AI has been called the “next industrial revolution,” a once-in-a-generation technological shift. But speak to people actually using these tools, and you’ll hear a different story. One of glitches, hallucinations, and unmet promises. So what’s really holding back adoption?

AI is rubbish

Let’s call it for what it is: a lot of AI is bad. It confidently invents facts. It can’t tell when it’s wrong. And it often produces content that ranges from bland to outright false.

The Future Skills Organisation says many Australian businesses are worried about the accuracy and integrity of AI outputs - particularly hallucinations - where AI tools generate information that simply isn’t true.

Data privacy is another red flag. Survey respondents across industries and demographics list it as their top concern. And older workers in particular say they just don’t trust AI enough to use it at work.

Workers are scared it will replace them

One of the biggest elephants in the room is jobs. According to a recent Pew Research Center survey, more than half of US workers who are aware of AI tools say they’re worried the technology could lead to workforce reductions.

For workers in sectors like customer service, content creation, or administration where AI is already being deployed, these fears are becoming very real. And when people are scared a tool will make them redundant, they’re unlikely to adopt it with open arms.

Regulatory limbo

Then there’s the law, or lack thereof. Research from KPMG, the University of Queensland and the AI Group shows that regulation (or the lack of it) is shaping AI adoption in surprising ways.

Businesses say they want guardrails. They’re asking for clearer rules around who is responsible when AI gets something wrong. They want a framework that spells out where and how AI can be used, and what rights workers and customers have.

Until those frameworks arrive (and they’re being debated in Australia and globally right now)many companies are choosing to wait.