The widespread adoption of artificial intelligence (AI) – particularly in “hidden” everyday applications – is creating a new and distinct form of digital inequality.

A team at the Hong Kong Baptist University,  led by communication researcher professor Sai Wang analysed more than 10,000 Americans’ engagement with AI.

It revealed that people with higher levels of education or income tend to be more aware of AI, more familiar with it, and more likely to use the burgeoning technology than those with a lower socioeconomic status (SES).

The researchers define AI awareness primarily as recognising the use of the technology in various context; familiarity, meanwhile, relates to people’s perceived knowledge of AI, regardless of their actual knowledge.

“Closing the AI awareness gap is essential, because if only people with higher income or education are aware of AI and its uses, this may reinforce social inequalities,” explained Wang. “It allows some groups to leverage advanced technologies for their advantage, while others are left behind.

“For example, job applicants who know that employers use AI for screening can better tailor their resumes, while those who lack this awareness might miss out on opportunities without realising it.”

Alongside its ability to empower individuals, AI also comes with the risk of harm, Wang noted.

“People with greater awareness may better understand both the opportunities and risks of AI – such as recognising and even creating deepfakes – while those with less awareness are more likely to be deceived or manipulated by these technologies,” she said.

In their study, Wang and colleagues analysed survey data on understanding of and attitudes towards AI collected from 10,087 US adults by the nationally representative American Trends Panel.

Past studies have suggested that wealthier and more educated people—alongside typically having more developed digital skills – are more likely to be encouraged to take advantage of AI tools, which in turn boosts confidence in using AI. These trends help explain why education and income emerged as significant predictors of AI usage in the current study.

That said, according to Wang, their study also revealed an unexpected finding: familiarity with AI was a stronger predictor of AI awareness than actually using AI.

“In other words, simply feeling knowledgeable or informed about AI was more closely linked to recognising where AI exists and how it is used compared to personally using AI technologies,” said Wang.

An explanation for this phenomenon may lie in how many common applications of AI have been so seamlessly integrated into the everyday apps and platforms of the digital lives that the addition is not obvious.

“For example, AI-driven recommendation systems on streaming platforms like Netflix or Spotify suggest content tailored to a person’s tastes,” says Wang. She continued: “Yet many users are unaware these are powered by AI and may see recommendations as random or neutral.”

In this way, the new digital inequalities being produced by AI are distinct from their predecessors.

“Traditional digital inequalities focus on access, skills/use, and outcomes—all of which tend to presume users are consciously engaging with technology,” continued Wang.

“However, AI is often built into everyday apps and platforms in ways users do not realise; many people interact with AI, such as through social media feeds or streaming recommendations, without knowing it.”

Because of this, the team explains, merely increasing access to AI-powered technologies may not be enough to close this awareness gap.

Instead, the researchers recommend indirect approaches to reduce this new digital inequality, in particular by familiarising people from lower SES backgrounds with key issues related to AI.

The team would like to see resources made available to increase engagement with AI-related topics, address public concerns and offer guidance on the ethical and responsible use of AI; basic AI concepts might also be integrated into educational curricula.

AI literacy programs, the researchers add, must include targeted guidance on how to identify “hidden” AI in daily life and understand its basic functions.

“It is imperative to work toward a more inclusive digital future in which technology empowers everyone and does not further marginalize any group,” the researchers concluded.