New research shows business uptake, public trust and fairness all lagging behind the narrative
Eighteen months into the generative AI boom, the gap between what the technology is promised to do and what it is actually doing appears to be widening rather than closing.
A clutch of new reports released this week, covering business adoption, consumer trust, career mobility and algorithmic fairness, suggest that, while AI is moving forward, it’s nowhere near as fast, evenly, or cleanly as the headlines suggest.
Massachusetts-based Daniel Research Group has updated its long-running forecast of the US personal devices market and estimates that only 18.4% of US businesses will be actively using AI to produce goods or services in 2026.
That is up from 11.2% in 2025 and just 3.9% in 2023, genuine growth, but the firm's analysis, built on the Census Bureau's Business Trends and Outlook Survey, found that activity remains concentrated among large companies and technology-heavy sectors, while small and mid-size businesses, which make up the bulk of US firms, are still on the sidelines.
Cost, a shortage of qualified staff, and unresolved questions about return on investment remain the biggest brakes on faster uptake, alongside what DRG describes as a deeper, recurring unease about handing decisions to machines.
Consumers are not buying the pitch
If businesses are cautious, consumers are openly skeptical.
A new Future of the Web survey from WordPress VIP, conducted by Talker Research among 2,000 respondents including 800 enterprise marketing decision-makers, found that 74% of people think the internet feels less human than it did a decade ago, and that the average person hits "bot fatigue" within about 40 minutes.
Six in 10 said AI showing up in a brand's messaging is actually a turnoff, and most could not name a single company they believe is using AI well in its marketing.
Search results or AI-generated answers without clear attribution are now the thing US consumers say they trust least online, ranking below medical bills, airline fees and confusing fine print. The vast majority, 86%, say they still click through to an original source after getting an AI summary.
However, enterprises are investing in the opposite direction with 60% of their audience reach now coming from third-party platforms they do not own, while only 17% plan to prioritize their own websites for growth next year.
What about jobs?
Where the technology is paying off most clearly is for individual workers, according to Randstad.
The talent firm's latest analysis of more than 35 million job postings since 2021 found that professionals who pick up advanced AI credentials are securing promotions up to 3.5 times faster than traditional peers, and that AI-integrated roles are commanding salary premiums of up to 25% over traditional equivalents.
Demand for distinctly human skills, creativity, emotional intelligence, ethical judgment, has also surged alongside it.
"While today's headlines often focus on how automation is leading to job displacement, our data reveals a powerful human judgment premium emerging across the US workforce," said Greg Dyer, CCO of Randstad North America. "The most successful professionals today are using AI to not only work faster, but also free up the headspace required for the critical thinking only a human can provide."
Bias built into the systems themselves
Perhaps the most uncomfortable finding comes from a peer-reviewed study led by Special Olympics in partnership with Oregon State University, published in the Disability and Health Journal.
Researchers asked five widely used large language models, including GPT-4-Turbo, to generate 25,000 short stories about everyday situations, comparing those describing a person with an intellectual disability against those that did not.
Across all five systems, stories about people with intellectual disabilities consistently portrayed them as more dependent, more childlike and more in need of supervision, with some stories more than 100 times as likely to carry a paternalistic tone.
"Artificial intelligence is quickly becoming foundational to how our world operates, from healthcare to education to economic opportunity. That makes this a pivotal moment. If we do not intentionally design for inclusion from the outset, we risk embedding bias into systems that will scale globally and persist for generations. We want to disrupt automated ableism from the start," said Nathan Cook, Special Olympics Chief Information and Technology Officer.
Overall, these studies, all released Tuesday, highlight some issues with AI that may prove to be disruptors to the disruptor! While AI clearly isn’t going away, its use may become increasingly divisive, adding a new metric to investment decision making.
