The next great software transition could widen access to technological power. But if AI arrives as a tool imposed by a few companies, at the expense of workers and communities, the public may reject the bargain. Open-source AI offers a different future—if we keep it viable.
A developer uses AI to build free alternatives to Adobe’s creative software. The early versions have rough edges. That is not the point. The point is that software once requiring specialized teams and years of work may increasingly be within reach of individuals and small communities. This is not just a story about competing with Adobe. It is a glimpse of a larger transition: the tools for making software becoming more widely available, just as the internet made publishing and distribution available to nearly everyone.
Every major transition like this creates new possibilities—and new concentrations of power. The internet lowered the cost of reaching an audience, enabling independent publishing, open collaboration, and businesses that could not have existed before. It also created new gatekeepers. Search, social media, app stores, cloud services, and advertising networks became powerful intermediaries. The lesson is not that openness defeats concentration, or that concentration defeats openness. It is that both can grow from the same technological upheaval.
AI is entering that unsettled phase now. Its trajectory will depend not only on what it can do, but on whether people accept the way it is being built and deployed—and whether the tools remain accessible beyond the companies that control the largest models and computing infrastructure.
Public skepticism is already visible. In a June 2026 Pew Research Center survey, 52% of U.S. adults said they were more concerned than excited about AI’s growing use in daily life, up from 37% in 2021. Only 9% said they were more excited than concerned. Among adults under 30, 55% were more concerned than excited. Yet this is not simple rejection: people continue to use AI even as their worries grow. That combination—adoption alongside distrust—is exactly what makes public acceptance consequential. (McClain & Park, 2026). Pew Research Center
The concern is not mysterious. People see AI changing work, creative life, education, relationships, and the information environment, while decisions about its use are often made far from those affected. They worry that employers will use it to cut jobs or intensify work; that creators’ material will be taken without consent; that automated systems will make consequential mistakes; and that the benefits will go to owners while the costs fall on workers and communities.
Those are not abstract fears. The International Energy Agency estimates that data centers used about 415 terawatt-hours of electricity in 2024, or 1.5% of global electricity consumption, and projects that figure could reach about 945 terawatt-hours by 2030. The agency emphasizes uncertainty in the forecast, and data centers serve more than AI alone. Still, their growing, geographically concentrated demand raises real questions about who pays for new power infrastructure and who bears its local effects. (International Energy Agency, 2025). iea.org
The danger is not that public anxiety automatically stops AI. It is that businesses and governments mistake deployment for legitimacy. A product can be adopted because it is bundled into workplace software, required by an employer, or offered as the only practical service—not because people trust the system or welcome the terms. If institutions press ahead while resisting meaningful rules on labor, data, copyright, safety, and environmental costs, they may turn skepticism into organized opposition.
That is where the labor history matters. The industrial transformation of the nineteenth century was not only a story of machines increasing production. It was also a struggle over who controlled work, who gained from productivity, and whether workers had power to shape the conditions imposed on them. The U.S. Department of Labor’s history of the industrial era describes workers confronting concentrated business power and building unions, cooperatives, reform groups, and political organizations in response (Montgomery, n.d.). The circumstances today are different; we should not pretend a new AI economy will reproduce the exact strikes and movements of the past. But the underlying conflict—technological change combined with unequal control and distribution—is familiar. dol.gov
The strongest response is not simply to tell people that AI will be useful. It is to show them that AI need not belong to a handful of companies. Open-source software already demonstrates how a shared technical foundation can support enormous economic activity. Hoffmann, Nagle, and Zhou (2024) estimated that widely used open-source software has a demand-side replacement value of $8.8 trillion to firms. That figure is an estimate based on a particular counterfactual, not a literal market valuation. Its broader point remains: software made freely available and improved by communities is already a major part of the global economy. ssrn.com
AI could extend that pattern. More capable open-weight models, cheaper inference, and AI-assisted programming can lower the barriers to building, adapting, and maintaining software. Stanford’s 2025 AI Index reported sharply falling costs for AI inference and narrowing performance gaps between open-weight and closed models on some benchmarks. These trends do not make the entire AI stack open: compute, chips, data, distribution, and the strongest models remain powerful choke points. But they make it increasingly plausible that useful capabilities can spread beyond a small group of providers. (Maslej et al., 2025). Stanford University
That possibility matters politically as much as technically. If people can run models locally, inspect and modify software, build alternatives, and share tools with others, AI becomes less like a service they must accept on someone else’s terms and more like a capability they can shape. Public education should make that distinction clear. “AI” is not one inevitable product or business model. The technology can be deployed as a centralized subscription controlled by a few firms—or as a more open ecosystem in which communities, businesses, educators, and individuals can make meaningful choices.
Open source is not a magic guarantee of fairness, safety, or equal access. It cannot by itself supply the energy, computing resources, public oversight, worker protections, or stable funding that responsible development requires. Nor should the label “open” be allowed to obscure systems that release model weights while keeping other important elements closed. But keeping open development viable preserves an alternative to dependence on a few providers. Without it, the public is left to choose among products controlled by the same concentrated interests that are asking for trust.
Even the people building advanced AI are publicly divided over what they are building and whether it can be controlled. In September 2026, Anthropic alignment researcher Evan Hubinger said he believed there was a greater-than-10% chance AI could kill all humans within a decade; other experts questioned the framing and the motives behind such warnings. That claim is a personal risk estimate, not an established scientific consensus. But the episode is still revealing: the people closest to the technology disagree not only about the scale of the danger, but about whether the current development race is responsible. When insiders themselves describe profound uncertainty, the public’s demand for accountability is not irrational. (Gerken, 2026; Vermeer et al., 2025).
AI will not live or die by public opinion in the narrow sense that a poll can halt a technology. It may spread regardless. But its ability to become a trusted, durable part of everyday life—and the form that adoption takes—depends on whether people see benefits they can share and institutions they can influence. A public that expects to lose work, pay the environmental costs, surrender creative material, and accept decisions made by unaccountable systems is unlikely to grant lasting consent simply because business leaders promise transformation.
The internet transition is happening again, but its outcome is not predetermined. We can let AI become another layer of concentrated infrastructure, with the public treated as a market to be captured and a workforce to be optimized. Or we can defend open tools, educate people about their practical potential, and insist that workers and communities have a say in how they are used.
The promise of AI is not just that it can make software faster. It is that more people may be able to make the tools they need. If we want the public to accept that future, we should make sure it can actually belong to them.
References
Gerken, T. (2026, September 9). Anthropic researcher believes more than 10% chance AI “could kill all humans”. BBC News. https://www.bbc.com/news/articles/ckgwy1k42w4o
Hoffmann, M., Nagle, F., & Zhou, Y. (2024). The value of open source software (Harvard Business School Strategy Unit Working Paper No. 24-038). Harvard Business School. https://doi.org/10.2139/ssrn.4693148
International Energy Agency. (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai
Maslej, N., Fattorini, L., Perrault, R., Gil, Y., Parli, V., Kariuki, N., Capstick, E., Reuel, A., Brynjolfsson, E., Etchemendy, J., Ligett, K., Lyons, T., Manyika, J., Niebles, J. C., Shoham, Y., Wald, R., Walsh, T., Hamrah, A., Santarlasci, L., … Oak, S. (2025). The AI Index 2025 annual report. Stanford Institute for Human-Centered Artificial Intelligence. https://doi.org/10.48550/arXiv.2504.07139
McClain, C., & Park, E. (2026, August 18). Young adults in the U.S. are increasingly wary of AI, concerned it will take jobs. Pew Research Center. https://www.pewresearch.org/short-reads/2026/08/18/young-adults-in-the-us-are-increasingly-wary-of-ai-concerned-it-will-take-jobs/
Montgomery, D. (n.d.). Chapter 3: Labor in the industrial era. U.S. Department of Labor. https://www.dol.gov/general/aboutdol/history/chapter3
Vermeer, M. J. D., Lathrop, E., & Moon, A. (2025). On the extinction risk from artificial intelligence (RR-A3034-1). RAND Corporation. https://doi.org/10.7249/RRA3034-1
