On June 30, the North Carolina Department of Information Technology announced the Statewide AI Strategy Roadmap created by the NC AI Leadership Council compiled by the Governor’s Office. Like many strategic plans, this plan paints an ambitious picture of where the state hopes to go in the coming years. It talks about responsible AI, workforce readiness, government modernization, education, privacy, and public trust.
These are all worthy goals, and perhaps more importantly, they recognize that artificial intelligence is progressing beyond just a technology issue. It’s rapidly becoming an economic development concern, a workforce concern, and increasingly a citizen concern. However, as I read through the roadmap, I realized that I was more interested in what was not included than what was included.
The document spends considerable time explaining what national institutions, educational institutions, employers, and communities should do to prepare for an AI-enabled future. However, there is little mention of the organizations developing the technology itself. This omission raises broader questions that reach far beyond North Carolina.
Who exactly is responsible for teaching society how to use artificial intelligence?
This question came to me because just the day before, June 29, California announced a statewide partnership with Anthropic, the company behind Claude. This agreement is notable not because California chose one AI provider over another, but because the arrangement reportedly extends beyond software license discounts. This includes training, technical assistance, and implementation support designed to help civil servants learn how to incorporate AI into their daily work.
In other words, California didn’t just buy the software. It purchased the ability.
This is not a subtle difference. This will significantly reshape the relationship between technology providers and government customers. I think this will be one of the defining questions of the AI era. For decades, we’ve thought of software as a product. Companies build applications, sell licenses, and sometimes provide documentation and technical support. In addition, success is highly dependent on the customer. When an organization fails to transform with a software product, it is seen as an implementation issue rather than a vendor issue.
For example, when the county purchased Microsoft Office 20 years ago, it was a software purchase. Employees might attend classes or watch tutorials, but no one expected Microsoft to partner with counties to reimagine permitting, budgeting, emergency management, and resident services. The value of software comes from the functionality within the application.
Enterprise software has traditionally been additive. The government purchased databases to manage records, spreadsheets to analyze budgets, and word processors to create reports. Each application made existing tasks somewhat faster or easier, but rarely changed the fundamental nature of the work.
Artificial intelligence is fundamentally different.
Unlike traditional software, AI is not just another application you install. It changes the way work is done, the way decisions are made, the flow of information, and increasingly the structure of the organization itself. Purchasing access to large-scale language models is becoming less like purchasing enterprise software and more like connecting to a new form of infrastructure. And infrastructure has always carried the expectations of broader society.
When roads were built across America, we didn’t just hand people the keys to their cars and wish them luck. We developed driver education, licensing systems, traffic laws, safety standards, and public investment to teach people how to operate safely within new transport networks. Governments at all levels are responsible for the effectiveness and safety of transportation systems.
Health care, another field rooted in the public interest, follows a different model. Much of the responsibility for safe implementation lies with private industry. Pharmaceutical companies conduct clinical trials, produce extensive documentation, educate physicians, monitor adverse events, and continue to support products long after they are on the market. Although governments establish regulatory frameworks, the private sector has substantial responsibility for ensuring that complex technologies are used appropriately.
Consumer-facing industries often take a different path and place much greater responsibility on individual users to achieve their own success. IKEA and Lego expect you to “build it yourself” using the included assembly instructions. KitchenAid doesn’t provide every stand mixer to cooking instructors, and Weber doesn’t ship every new grill home to pitmasters. DeWalt will not send a carpenter to your home after you purchase a new table saw. Apple and Samsung don’t spend resources teaching you how to get the most out of your smartphone.
Consumer companies provide manuals, safety information, and customer support, but the responsibility for achieving success lies with the owners. The same applies to liability for product misuse.
In most cases, the buyer beware strategy works well for society. After all, toys, household appliances, tools and furniture are not products that often intersect with “public interest”. Platform technologies such as smartphones are more of a gray area. Neither cell phone manufacturers nor the software app providers that utilize cell phones are required to provide extensive public education about the risks of covert collection of personal data, disinformation, distracted driving, phishing attacks, and more, despite all of these phone-related challenges to society.
It’s worth discussing when product manufacturers should be held more accountable than they currently are. The firearms industry is another illustrative example. Manufacturers are not required to provide extensive public education on gun use any more than we provide driver training in public schools. We also do not provide certification or licensing similar to the DMV. But gun misuse creates all sorts of challenges and responsibilities for society to manage. Public safety is the basis of the public interest.
Private industry, government, and consumer responsibility are three different approaches to answering the same question. When technology has a significant impact on society, who is responsible for helping people make good use of it??Artificial intelligence doesn’t answer that question.
Most discussions about responsible AI naturally focus on privacy, bias, copyright, cybersecurity, or model safety. These are important conversations. But they all assume that organizations already know how to responsibly implement AI in the first place. I would argue that most people don’t. And this is especially true for local governments.
It’s easy to imagine large state agencies building internal AI teams, hiring experts, and developing governance frameworks. It’s much harder to imagine a rural county with a smaller IT budget doing the same thing. Many municipalities in North Carolina struggle to hire technical staff, even if they have the budget. Few companies will have agile engineers, AI governance specialists, and a dedicated implementation team at their fingertips. But these may be the very communities that stand to benefit most.
Artificial intelligence offers the potential to automate administrative tasks, simplify permitting, assist with grant writing, improve citizen communication, accelerate planning, modernize procurement, and make government services more accessible without significantly increasing headcount. For small communities that have long operated with limited resources, these productivity gains can be transformative. But only if someone shows you how.
Policymakers have long focused on closing the digital divide by expanding broadband. That investment was necessary but premature. But as connectivity becomes more pervasive, another divide is emerging that has little to do with fiber optic cables or wireless coverage.
The next dividing line may be competency.
Even though counties have broadband, cloud infrastructure, affordable AI subscriptions, and modern computers, they may still feel left behind because no one has developed the institutional knowledge needed to redesign government around these new capabilities. It’s not just a technical challenge. It’s a workforce challenge. It is also an issue of economic development.
North Carolina has invested heavily in attracting data centers, semiconductor manufacturing, biotechnology, and advanced industries. These investments create jobs, tax revenue and long-term economic opportunities. But if we stop thinking about infrastructure after the building has been constructed, we risk repeating the mistakes we should already be familiar with.
For decades, we relied primarily on the private sector to expand broadband access. The economy worked well in densely populated areas. In many rural areas, they work much less effectively. As a result, some of our states have spent years on the bad side of the digital economy (and some still do). This wasn’t because the technology wasn’t valuable, but because incentives didn’t always align with universal access.
I have previously referred to these regions as the “digital rust belt.” In these regions, economies are largely failing because the necessary technological infrastructure either does not yet exist or is underperforming. And even where connectivity was available, there were insufficient programs to train employees on how to best utilize it for e-commerce, distance learning, telemedicine, and other benefits.
Artificial intelligence presents an opportunity to avoid deepening the digital rust belt. But this requires rethinking what we expect from the companies that build foundational AI models. If these organizations increasingly describe their technology as a general-purpose infrastructure that can transform any industry, I argue that it should be measured not just by benchmark scores, inference power, and subscription price, but also by how responsibly and effectively society can use that technology.
Imagine what would happen if partnerships like California’s became the norm rather than the exception. What if all AI deployments across the state included structured workforce development? What if enterprise AI bundled and contracted implementation coaching, ethics training, governance templates, and continuing education along with the software itself? What if AI companies competed not only on the performance of their models but also on the success of the communities they helped modernize? That would profoundly change the relationship between technology companies and society.
Government will continue to play an important role. Public education, standards, procurement policies, and accountability remain within the responsibility of the public sector. However, it does not seem unreasonable to expect AI providers to actively participate in building an AI-enabled workforce. In fact, this could be one of the defining competitive advantages of the next decade.
North Carolina’s roadmap establishes an important vision. It identifies a number of pertinent issues and sets out a direction that deserves widespread support. The more difficult work begins now.
As nations move from strategy to implementation, the next question should not simply be how governments deploy artificial intelligence, but how governments and industry can work together to build an AI-savvy society, starting with the AI-enabled public sector itself.
We shouldn’t make the same mistakes when it comes to AI as we did with broadband. Building the infrastructure is only half the job. The other half is building the human capacity to use it. Whether that responsibility ultimately lies with government, industry, individuals, or some combination thereof, we must answer that question by design, not by chance.
We have to get this right. I believe it’s as important to train people how to use AI in ways that protect their data, privacy, and digital security as I believe it’s important to learn to drive a car safely. Or to take medicine. Or to handle guns safely.
After all, the future of artificial intelligence will not only be determined by the quality of the models we build. It depends on the ability of the person using it. If AI is destined to become infrastructure on the scale of power, highways, healthcare, and the internet, teaching society how to use it safely and effectively is not an optional feature. It’s part of the infrastructure itself.
