A new study by researchers at the University of Toronto, University of Oklahoma, the Census Bureau and the National Bureau of Economic Research Stanford examines the impact of artificial intelligence in American manufacturing. Utilizing data from the U.S. Census Bureau for the years 2017 and 2021, the researchers identify a productivity pattern known as the J-curve. This indicates that companies experience short-term productivity losses before realizing long-term gains following AI adoption. The study highlights that initial use of industrial AI leads to significant increases in work-in-progress inventory and investment in robotics, while also resulting in labor reductions and declines in profitability. Notably, older businesses are more affected by these losses, though growth-oriented strategies can mitigate them. The findings suggest that early adopters of AI see stronger growth between 2017 and 2021, provided they survive the initial phase. The research emphasizes that AI’s benefits may not be immediately evident, particularly for established firms.
A panel of artificial intelligence researchers has offered that the field is currently pursuing the development of human-like intelligence in the wrong manner. This insight emerged from the Association for the Advancement of Artificial Intelligence’s 2025 Presidential Panel on the Future of AI Research, which included a report from twenty-four experts. The report noted that seventy-nine percent of respondents believe that public perceptions of AI capabilities do not align with the reality of research, with ninety percent stating that this mismatch is hindering progress. Notably, seventy-six percent of surveyed researchers indicated that merely scaling up current AI approaches will not lead to achieving human-like intelligence. The panel emphasizes a cautious and collaborative approach to AI development, advocating for safety and ethical governance.
A recent Pew Research Center survey reveals a significant gap in perceptions about artificial intelligence between experts and the general public. While fifty-six percent of AI experts believe that AI will positively impact the United States over the next twenty years, only seventeen percent of adults share this optimism, with thirty-five percent fearing a negative impact. The survey, which included over five thousand randomly selected adults, indicates that most Americans are less confident about AI improving their workplaces, healthcare, and education. Notably, sixty-four percent of the public expects fewer jobs due to AI, compared to just nineteen percent of experts. Despite the rapid rise of generative AI tools, many Americans express more concern than excitement, particularly regarding issues like job loss and misinformation. The report underscores the desire for more personal control over AI and highlights widespread worries about government oversight.
Why do we care?
The University of Toronto study highlights a crucial lesson for any business considering AI: initial productivity declines are not uncommon. For manufacturers—and likely other sectors—the adoption phase brings complexity, increased investment, and even reductions in profitability before yielding gains. Older and more established businesses are hit harder, often due to legacy processes and cultural resistance. Growth-oriented strategies can mitigate these losses, emphasizing the need for well-planned AI implementation rather than a quick fix.
The key takeaway here is the need for a balanced approach. Providers should avoid hyping AI solutions without evidence and instead focus on practical, tangible applications that clients can realistically benefit from. Moreover, staying informed about AI research can help maintain credibility and manage expectations.