by Anonymous
San Francisco has long been a city where technological change arrives before the rest of the country. The latest transformation is artificial intelligence. AI is creating enormous wealth and attracting investment, but it is also beginning to change the labor market in ways that could have profound consequences for housing and homelessness.
The central question is no longer simply whether artificial intelligence will eliminate jobs. The more important question is: What happens to people when their jobs disappear faster than they can find new ones—and when housing remains unaffordable?
For San Francisco, that question should be treated as a homelessness issue as much as an economic one.
The AI employment shock is already visible
It would be premature to claim that AI has already caused mass unemployment. Current evidence does not support that conclusion. California’s AI workforce tracker, launched in 2026, says there is not yet evidence of large-scale AI-related layoffs statewide. At the same time, however, the state has found increased unemployment claims among workers in highly AI-exposed occupations, particularly in the Bay Area.
The San Francisco metropolitan area is particularly exposed.
A 2026 analysis of local employment found that approximately 21% of Bay Area jobs have an AI exposure share above 50%, compared with 18% nationally. Nearly 70,000 people in the region work as software developers, an occupation with an AI exposure score of approximately 45%. Office clerks, cashiers, retail workers, managers and other occupations are also substantially exposed.
That does not mean 21% of San Francisco workers are about to lose their jobs.
It does mean that a significant portion of the local economy is vulnerable to technological restructuring.
And restructuring does not have to mean an immediate mass layoff to create hardship.
AI can reduce the number of new employees companies need. It can eliminate entry-level positions. It can allow one employee to accomplish the work previously performed by several people. It can reduce wages and bargaining power even when workers technically remain employed.
Recent research has already raised concerns about declining opportunities for younger workers in AI-exposed occupations. The San Francisco Federal Reserve is also examining how AI is affecting employment, wages and workers throughout the region.
This creates a particularly dangerous possibility for people who are already living paycheck to paycheck. Losing a job can become losing a home.
Homelessness rarely begins with someone simply deciding to live on the street. For many people, homelessness is the final stage of an economic chain: job loss → missed rent → eviction → temporary housing → exhaustion of savings → homelessness.
If AI accelerates job displacement while housing costs remain high, that chain could move faster.
San Francisco is especially vulnerable because housing costs are already extraordinarily sensitive to economic conditions. Recent reports found asking rents for vacant apartments increased 14% between March and July 2026, while the city’s AI-driven economic boom has contributed to rising housing demand. At the same time, tech employment growth has remained relatively modest because AI is both creating and eliminating jobs.
That creates an economic contradiction.
AI can make San Francisco wealthier while simultaneously making it harder for displaced workers to afford San Francisco.
The people most likely to benefit from the AI boom may be the people who own companies, equity, intellectual property, infrastructure and highly specialized skills.
The people most vulnerable may be workers whose jobs become less valuable.
If that distribution continues, San Francisco could experience an unusual phenomenon: economic growth alongside growing economic insecurity.
The danger of treating homelessness primarily as a street problem
This brings us to San Francisco’s current homelessness strategy under Mayor Daniel Lurie.
The administration deserves credit for recognizing that the existing homelessness system needs reform. Lurie’s March 2025 directive called for expanding interim housing and treatment capacity, improving case management, increasing accountability for service providers and strengthening workforce-development programs.
Those goals are reasonable.
The problem is what happens when homelessness policy becomes primarily focused on removing visible homelessness from public spaces rather than preventing people from becoming homeless in the first place.
The city has increasingly emphasized encampment enforcement, treatment, sobriety-oriented programs and specialized facilities. Lurie’s administration has also advanced the RESET Center, which is intended to provide an alternative to jail or hospitalization for people arrested for public intoxication and connect them with treatment.
There is an important distinction here:
Addressing dangerous behavior is not the same thing as addressing homelessness. A person can be moved from one sidewalk to another without becoming housed, can be arrested without becoming economically stable, can be offered treatment without receiving permanent housing. And a tent can disappear from a neighborhood without the underlying causes of homelessness disappearing.
That distinction matters enormously if AI begins producing a larger population of economically displaced workers.
San Francisco cannot enforce its way out of an economic crisis.The Supreme Court’s 2024 decision in City of Grants Pass v. Johnson gave municipalities considerably more authority to enforce laws against public camping.That legal authority, however, does not make enforcement a homelessness solution.If the number of people losing housing grows faster than the number of genuinely accessible housing opportunities, enforcement can become a revolving door.
Someone is removed from an encampment.
They move somewhere else.
They are removed again.
Eventually they may receive a citation, an arrest, a shelter placement, a temporary bed or some other intervention.But if they still cannot afford permanent housing, the underlying problem remains.This is particularly troubling in an economy where technological displacement could affect people who have never previously experienced homelessness.The future homeless population may not look exactly like the homeless population of the past.It could include former office workers, administrative employees, customer-service representatives, junior programmers, designers, analysts and other workers who suddenly discover that their previous career path has disappeared.
Shelter is necessary—but prevention is even more important.San Francisco’s shelter system is also undergoing significant changes.
The Lurie administration originally promised to add 1,500 shelter beds within six months. By April 2026, reporting found that 863 new beds had opened, but closures meant the net increase was considerably smaller. The administration subsequently shifted its emphasis toward providing the “right” kinds of beds, including treatment and stabilization facilities.
Specialized treatment facilities can be valuable, but San Francisco also needs something much less dramatic: a massive homelessness-prevention system. If a worker loses a job because their position was automated, the city should not wait until that person is sleeping outside before intervening. Imagine instead a system where someone who loses an AI-exposed job can immediately access emergency rental assistance, eviction prevention, temporary income support, workforce retraining, free AI and technology education, job placement, mental-health and substance-use services when appropriate, transitional housing, and eventually permanent affordable housing.
That is a homelessness strategy designed for the economy of the future rather than the economy of the past.
AI policy must become housing policy.
California has already begun considering how labor protections should change in response to AI. The state is examining whether existing worker-notification laws adequately address technological displacement, while proposed legislation would create additional notice requirements for certain AI-driven layoffs.
San Francisco should think even further ahead.
If AI dramatically increases productivity, the benefits cannot simply accrue to companies and investors while workers absorb the economic risk. The city should be asking: Who receives the benefits of automation? Who pays the cost when jobs disappear? What happens to workers during the transition? How do we prevent technological unemployment from becoming homelessness?
Those are not anti-technology questions. They are questions about whether technological progress benefits an entire city or primarily the people who own the technology.
The real test of San Francisco’s homelessness policy
Mayor Lurie’s administration has emphasized accountability, measurable outcomes and reforming a system that has consumed enormous public resources without eliminating homelessness. That concern is legitimate.
But accountability should not simply mean measuring how many people were removed from sidewalks.It should mean measuring how many people remained housed. How many people avoided eviction? How many people obtained permanent housing? How many people obtained employment after losing their previous jobs? How many people successfully transitioned from temporary assistance to independence? And, most importantly, how many people never became homeless in the first place?
That last number may become the most important measurement of all.
San Francisco has a choice. Artificial intelligence could produce enormous prosperity, or it could produce a society in which fewer workers are needed to produce more wealth. Those two outcomes are not mutually exclusive, but the danger is assuming that economic growth automatically protects everyone.
It does not.
If AI makes companies more productive while reducing the demand for human labor, San Francisco will need to rethink the relationship between employment, income and housing. Otherwise, the city could find itself with booming technology companies, rising rents, increasing wealth—and a growing population of people who simply cannot find a place in the new economy. That is why homelessness policy cannot be limited to what happens after someone ends up on the street.
The homelessness crisis of the AI era must be addressed before people lose their homes. San Francisco should prepare for technological displacement by investing aggressively in affordable housing, eviction prevention, retraining, income stabilization and pathways into new careers.
Enforcement may be necessary in certain circumstances. Treatment may be necessary. Shelter is necessary. But none of those things substitute for the fundamental goal: keeping people housed. If San Francisco waits until AI-displaced workers are sleeping in tents before helping them, the city will already have failed.
The smartest homelessness policy for an AI-powered economy is not simply to move people off the streets. It is to make sure they never have to live there in the first place.
The author requested this piece to be published anonymously.
