Goldman Sachs estimates that 300 million jobs globally will be affected by AI. McKinsey reports that 72% of companies are already using AI in at least one business function. Global AI investment has exceeded $200 billion, according to Stanford's AI Index. And yet, 24 million Americans still lack broadband access, per FCC data. The conversation about AI has been dominated by one question: which jobs will disappear? That is the wrong question. The more important one is: what is AI actually revealing about the systems we already have?
AI is not arriving into a world of well-functioning systems and disrupting them. It is arriving into a world of fragile systems and exposing them. Customer service, data entry, content production, scheduling, intake processing, basic analysis -- these were never strong systems. They were labor-intensive workarounds. They required human beings to perform repetitive, low-judgment tasks at scale because nobody built the infrastructure to handle those tasks any other way. AI is not replacing excellent processes. It is replacing the absence of process.
The Fragility Was Already There
Consider customer service. Most large companies run customer support through call centers staffed by people reading scripts, navigating decision trees, and looking up information in databases they can barely search. The system was never designed for quality. It was designed for volume. The customer experience has been mediocre for decades -- long hold times, repeated information requests, inconsistent answers. AI chatbots are not disrupting a great system. They are replacing a bad one that companies tolerated because the labor was cheap enough to sustain it.
Data entry is another example. Organizations across healthcare, government, logistics, and finance employ thousands of people to manually transfer information from one system to another. Not because human judgment is needed for that transfer, but because the systems were never integrated. The data entry role exists to compensate for architectural failure. When AI automates that role, the job disappears, but the underlying problem -- that two systems should have been connected from the start -- remains visible for the first time.
Content production follows the same pattern. The explosion of content marketing over the last decade created a massive demand for written material that was often formulaic, lightly researched, and optimized for search engines rather than readers. AI can produce that content faster and cheaper because the bar was already set at volume over quality. The work that is most vulnerable to AI is the work that was already closest to mechanical.
AI is not disrupting strong systems. It is making the weakness of existing ones impossible to ignore.
The Digital Access Gap Compounds Everything
The cities and communities that were already struggling with basic digital infrastructure are about to fall further behind. When 24 million Americans lack broadband -- and tens of millions more have connections too slow or unreliable for modern applications -- AI does not arrive as an equalizer. It arrives as an accelerant of existing inequality.
In Detroit, large sections of the city still lack reliable high-speed internet. The digital literacy gap is significant. Small businesses operate without basic digital tools -- no CRM, no analytics, no automated scheduling. These are not businesses that are going to adopt AI and leapfrog their competitors. These are businesses that are still trying to get a functional website. The AI conversation happening in Silicon Valley boardrooms might as well be on a different planet.
This is not unique to Detroit. Rural communities across Appalachia, the Delta, and the Great Plains face the same access problem. Mid-sized cities like Memphis, Flint, and Youngstown have populations that are digitally underserved. The infrastructure gap that existed before AI is now the gap that determines who benefits from AI and who gets displaced by it.
The risk is not just that some people lose jobs to AI. The risk is that the communities least equipped to adapt are the ones most affected by the transition. And the current policy response -- workforce retraining programs, which we have already discussed the limitations of -- is nowhere near sufficient for what is coming.
Shifting the Conversation
The public discourse around AI has been stuck in a prediction loop. Which jobs will be automated? How many millions will be displaced? When will it happen? These are relevant questions, but they keep the conversation focused on outputs -- job counts, role categories, timeline projections -- rather than on the underlying structural issues that AI is surfacing.
The more productive conversation is about systems. If AI can replace a customer service operation, what does that tell us about how that operation was designed? If AI can automate data entry across an entire organization, what does that reveal about the integration failures in that organization's technology stack? If AI can generate marketing content indistinguishable from what a team of writers was producing, what does that say about the content strategy?
In each case, AI is not the problem. AI is the diagnostic. It is showing us, with uncomfortable clarity, which systems were built on manual labor as a substitute for actual design.
The question is not "what jobs will AI take?" The question is "what systems were we propping up with labor that should have been replaced by design?"
What Needs Rebuilding
If we accept that AI is revealing fragility rather than creating it, the response changes. Instead of defending jobs that existed because systems were poorly designed, we should be rebuilding the systems themselves. That means different things in different sectors.
In government, it means finally integrating the patchwork of disconnected databases, intake systems, and case management platforms that force residents to provide the same information to five different agencies. AI can help here -- not by replacing caseworkers, but by connecting the systems those caseworkers are forced to navigate manually.
In healthcare, it means addressing the administrative burden that consumes roughly 30% of healthcare spending. Prior authorizations, claims processing, scheduling, and record management are ripe for automation -- not because the people doing those jobs are unnecessary, but because the processes themselves are unnecessarily complex. Simplify the system, and the role of the human in it changes.
In education, it means rethinking how we prepare people for work in a world where the baseline skills are shifting. Digital literacy is no longer optional. The ability to work with AI tools is becoming as fundamental as the ability to use a spreadsheet was twenty years ago. Schools and workforce programs that are not incorporating this are already behind.
In small business, it means building the infrastructure layer that allows a three-person company to access the same operational capabilities as a three-hundred-person company. AI can be the great equalizer for small businesses -- but only if the access, education, and support systems exist to make it usable.
Where This Goes Next
The real disruption from AI is not automation. It is visibility. For the first time, the structural fragility of systems that have been propped up by manual labor is fully exposed. The organizations, cities, and economies that respond by rebuilding those systems -- with better architecture, better integration, better design -- will come out ahead. The ones that respond by trying to protect the old structures will fall further behind.
The next piece in this series examines one of the most visible examples of systemic fragility in America: housing. The national narrative says we do not have enough homes. The data tells a more complicated story -- one that has less to do with supply and more to do with coordination.
