Artificial Intelligence: What It Offers, What It Costs, and How to Respond

AI
Written By Don Bradford
Artificial Intelligence: What It Offers, What It Costs, and How to Respond

Artificial Intelligence: What It Offers, What It Costs, and How to Respond

LodiEye — July 2026

Overview

Artificial intelligence has become a strategic asset, as important to national power as past leaps in weapons and aviation. It affects military strength, intelligence, and the economy. China and Russia are pouring money into it and racing to put it to use. Washington now treats falling behind as a national-security risk in its own right. That makes staying ahead a real priority, not a talking point.

The costs, though, are just as real, and they land close to home. The benefits of this technology are measurable and growing. So are the burdens: higher electricity bills, heavy water use, and pressure on jobs. Those burdens fall hardest on the towns that host the buildings. Public worry has run ahead of the harm measured so far, but it senses correctly where things are going. That worry deserves a real answer, not a brush-off.

Here is the challenge this report takes up: how can the United States stay ahead in military and technical strength while shaping its AI spending at home to address honest concerns about the environment, the economy, and jobs?

People often treat these as a trade-off: back off to protect communities, or push ahead and let the costs fall where they may. Neither works. Backing off hands the lead to rivals who face no such limits. Pushing ahead without limits lets the local harms grow. The task is to do both at once.

The answer is to steer, not stop. Keep investing where the country gains a real edge, and aim rules and public money straight at the harms people can actually see. Competitiveness and community well-being go together; they are not a choice between the two. This report sorts the proven benefits from the real risks and the exaggerated fears, explains why public worry raises the stakes, walks through the rules taking shape, and ends with a live local test of the challenge: the data-center debate in Lodi.

1. What the technology has actually delivered

The strongest case for these tools rests on results you can measure, not promises. A blunt clampdown would throw these gains away along with the risks.

  • Medical care. These tools have matched or beaten doctors at reading certain scans, such as mammograms, catching diseases like breast cancer earlier. They have also sped up the hunt for new drugs.
  • Fraud and security. Banks use them to catch fraud as it happens, blocking more real fraud while flagging fewer honest purchases by mistake, and to spot fake identities older systems missed.
  • Everyday work. In business and government, they handle repetitive paperwork, cut mistakes, and make sense of large piles of data, which lowers costs and improves decisions.
  • Schools. Used carefully, they cut the busywork that eats up teachers' time and help tailor lessons to each student. They support teachers rather than replace them.

The pattern is the same everywhere: these tools work best when they help a trained person do a specific job, and worst when they are left to run on their own.

2. The concerns worth taking seriously

The concerns below are backed by real-world evidence, not guesswork. These are the problems that sensible rules and spending should target. They come in three kinds: problems inside the tools, the strain data centers put on host towns, and lost jobs.

2a. Problems inside the tools

  • Bias. Because these tools learn from past data, they can repeat human bias in hiring, lending, and medical care.
  • Privacy. They need huge amounts of data, which raises the risk of leaks and misuse, especially with sensitive records.
  • Security holes. They can be tricked or fed bad data to make them misbehave.
  • Confident errors. They sometimes state false things with complete confidence, which is dangerous in medicine or law.
  • Hard to explain. It is often hard to say why one of these tools reached a decision, which makes it hard to challenge.
  • Unwatched use. They spread inside organizations without oversight, and they get worse over time if no one checks them.

2b. The burden on host towns

This is where public worry stands on the firmest ground, because the costs are physical and land on the people who live nearby.

How Americans view data centers, by impact area

Source: Pew Research Center survey of 8,500+ U.S. adults, January 2026.

  • Huge power use. These buildings use enormous amounts of electricity and water. By 2034, the data centers running this technology are expected to use about as much electricity as all of India, a country of more than 1.5 billion people.
  • Strain on the grid, higher bills. New centers push against the limits of the local power supply and can raise electricity and water prices. That is why 38% of Americans say data centers are bad for home energy costs, against just 6% who see a benefit.
  • Where they cluster. The country already has more than 4,000 data centers, most of them in Virginia, Texas, and California. That puts our state in the middle of the fights now breaking out.
  • Quality of life. Nationally, 30% of people say data centers hurt the quality of life for nearby residents, against 6% who say they help.
  • A mixed picture. They are not all cost. They bring building jobs, permanent jobs, and tax money, which is why 25% of Americans call them good for local jobs and 23% good for local taxes. But an unchecked building boom could still drag on the wider economy while concentrating wealth.

2c. Lost jobs

This is where public fear runs furthest ahead of what has actually happened, though the early warning signs are real.

  • The measured loss is still small. Job losses tied directly to this technology have stayed below 15% in recent years, even though many workers believe the damage is already far worse.
  • Young workers feel it first. Yale researchers find the clearest losses among recent college graduates, making that first job harder to get.
  • Uneven and sometimes hidden. Adoption at work is patchy, and some employers have brought these tools in without telling their workers.
  • The big effects are still ahead. The nonpartisan Congressional Budget Office calls the effect on growth, jobs, and wages a possibility, not something that has already arrived.
  • Fear ahead of the facts. In Britain, six in ten expect the technology to kill more jobs than it creates, reacting to forecasts rather than the actual numbers.

3. Public worry and why it raises the stakes

Public worry is running ahead of the measured harm, though it senses correctly where things are headed. A January 2026 Pew survey of more than 8,500 Americans found 39% believe data centers are mostly bad for the environment, against 4% good, and 38% bad for home energy costs. Opinion turns positive on local jobs and taxes. The worry crosses party lines but is not even: half of Democrats call data centers bad for the environment, against 31% of Republicans. And the more people learn about these buildings, the more negative they become.

Question What people believe What the evidence shows
Environment Mostly bad (39%) Real and growing; huge power use projected by 2034
Home energy costs Mostly bad (38%) Genuine strain on the grid and local prices
Jobs Many expect heavy losses Losses so far under 15%; young workers hit first
Local jobs and taxes Modestly positive Real, but only in a few places
The wider economy Broad worry Big effects still ahead

Here is why this matters. Job fears are about what might happen, while the environmental costs are visible right now. So public anger tends to be broad rather than aimed. Broad anger produces blunt answers, like bans and moratoriums, instead of the specific protections that would actually help residents. The better path turns real worry into clear, checkable demands: protect the water and the power grid, help workers who lose jobs, and require honest disclosure.

4. The rules taking shape

Rules are taking shape unevenly. Europe passed one sweeping law. The United States has mostly left it to the states after stepping back at the federal level.

  • Europe. The EU's law, phasing in through August 2026, sorts uses by risk, bans the most dangerous, and puts strict duties on risky uses like hiring and lending. Fines can reach €35 million or 7% of a company's global sales.
  • Federal U.S. There is no sweeping federal law. A 2023 order requiring safety testing was revoked in January 2025 in favor of fewer rules.
  • California. A 2025 law (SB 53) makes the biggest developers publish safety plans, report serious incidents within 15 days, and protect whistleblowers. A separate law effective January 2026 (SB 942) requires that computer-made content be labeled.
  • Other states. Colorado and New York City now require bias testing and impact reviews for risky uses.

Despite different approaches, nearly all these rules ask for the same four things: sort uses by risk, be open about how the tools work, guard against bias, and make clear who is responsible. The lesson is that targeted rules can fix the real harms without choking off useful work.

5. National security and staying ahead

The overview explained why the stakes are so high. Here is the detail behind it. A leading Harvard analysis points to three areas where this technology shifts national power.

  • Military strength. It creates new capabilities and makes existing ones cheap enough that even weak states and armed groups can reach far with precision.
  • Information. It sharpens spying and analysis, but it also makes convincing fakes cheaper and better, which eats away at public trust.
  • The economy. It could touch off a shift in which a country's population size matters less to its power, letting smaller countries with an edge punch above their weight.

As one MIT expert puts it, modern security depends on gathering and making sense of as much information as possible, and this technology boosts both. That race carries its own danger. Rushing untested systems into the field can cause accidents. The technology opens new targets for attack, and it makes hacking cheaper and harder to stop. The sensible national course does three things at once: keep the lead, encourage peaceful uses, and manage the worst risks, through research funding, work with allies, limits on selling sensitive tech to rivals, and firm protection of civil liberties.

6. What to do about it

People often treat national strength and local well-being as opposites. They don't have to be.

Approach Effect on the country Effect on communities
Block everything Hands the lead to rivals Local harms stay; the money to fix them dries up
Allow anything Keeps the lead but invites accidents Power, water, and job harms grow
Steer it wisely Keeps useful investment Aims rules and money at power, water, and workers

For any organization or city, steering it wisely comes down to a few concrete steps:

  • Use a known safety playbook. Adopt the federal NIST framework or the ISO 42001 standard, the same ones state and European rules point to, so oversight targets real problems without blocking useful work.
  • Keep track of the tools. Set up a small group across departments to log where these tools are used and who is responsible for each one.
  • Check them regularly. Test for bias, security holes, and confident errors.
  • Prove the value. Measure results before and after, and expand only what earns its keep.
  • Invest at home. Put money into retraining workers and into protecting the power grid and water supply.
  • For residents. Turn real worries into specific, trackable demands rather than open-ended opposition that pushes leaders toward a blanket ban.

7. The Lodi data-center debate

Lodi is not watching this debate from a distance. A proposal to look into a data center became a real local fight, and that is exactly why the rest of this report matters here. Lodi has faced the same choice the country faces.

How the fight started

At the May 6, 2025 council meeting, then-Councilman Cameron Bregman asked whether Lodi should look into data centers. He called it worth a serious look, though he never put forward a detailed plan. He also backed the Lodi Energy Center Hydrogen Project as a way to "keep energy and utility costs low for residents." The two ideas are separate, but they share the White Slough site. The question split city leaders. As the Lodi News-Sentinel put it: "Councilman Cameron Bregman says yes. Mayor Ramon Yepez and others say no."

The Mayor's stand

Ramon Yepez joined the council in March 2023 as the first Latino councilmember in the city's history, won re-election in 2024, and became Mayor in January 2026. He opposes the idea, and he put it plainly: "A hyperscale data center is the wrong project for Lodi." His worry is that a power-hungry facility would strain the city's grid and push up residents' utility bills. The national evidence backs him: 38% of Americans already call data centers bad for home energy costs, and the strain on local prices is well documented.

A statewide wave of bans

Lodi's fight sits inside a bigger pattern. In June 2026 alone, Indio passed a 45-day moratorium, Monterey Park moved to become the first U.S. city to permanently ban data centers by public vote, and Imperial County's approval landed in court over water. Nearby Oakley had already dropped its own data-center plans in March 2026 after residents pushed back. These are the blunt rejections that follow when worry hardens into a flat no.

Why Lodi could handle this well

Lodi holds some cards most towns don't, which could let it set firm terms instead of just saying yes or no.

  • Its own power company. Lodi Electric has served the city since 1910 and covers about 27,400 accounts. Its residential rate, around 19.6 cents per kilowatt-hour, is among the lowest in the region and roughly 56% below neighboring PG&E's rate of about 44.8 cents. Because the city owns the utility, it can set the terms for a big new customer and protect existing residents, which is exactly the Mayor's concern.
  • Recycled water. The White Slough plant has treated the city's wastewater since the 1920s and already supplies recycled water for power generation. It could serve a heavy user without draining the drinking-water supply, which answers the water fight playing out elsewhere.
  • On-site power, with cautions. A hydrogen-blending project at White Slough uses that recycled water at no cost to local drinking water. But hydrogen blends are safe only in small amounts, can raise emissions in older equipment, and cost more, so any expansion needs firm, public conditions.
Local factor The advantage What to watch
Power rates Among the lowest in the region; city-owned utility Protect residents from a big new load, the Mayor's main worry
Water Recycled water at no cost to drinking supply Delta communities flagged as at risk
Power on site Includes a hydrogen-blend project Safety limits, emissions, and cost of hydrogen
Jobs Data-center revenue plus retraining grants Farm and seasonal workers exposed to change
City government County is an early, careful adopter No formal City of Lodi policy yet

Where Lodi landed, for now

Lodi didn't say yes, and it didn't say no. Based on the public record, the council set the data-center question aside. There is no ordinance banning data centers, and there is no plan to study one further. The idea is on the shelf.

That pause has an upside. Lodi skipped the outright bans that Monterey Park and Oakley reached for, so its options are still open. If a real proposal ever comes, the city keeps the leverage its neighbors gave up.

But a pause is not a plan. The homework the careful path needs, what a big new customer would do to local power rates, how much recycled water it could use, what conditions the city would demand, still hasn't been done. Lodi's real advantages, its own utility and its recycled water, only matter if the city knows its own numbers before a developer shows up with theirs.

So the point for Lodi is simple. Tabling the question buys time. It does not answer it. The city can use this quiet stretch to do the groundwork now, on its own terms, or it can wait and be forced to decide in a hurry later, under pressure, which is how communities end up reaching for a ban.

LodiEye is the original civic-reporting and analysis arm of Lodi411.com, a citizen-run civic data and transparency platform serving Lodi, California and San Joaquin County. LodiEye gathers information of public interest, applies editorial judgment to public records, meetings, and data, and publishes original explanatory reporting for its readers — the work of a newsroom, and a representative of the news media as that term is defined under federal law. Our reporting emphasizes primary sources, public data, and full source transparency so readers can check every claim. LodiEye complements, and does not replace, the other outlets covering this region; for additional reporting on Lodi, San Joaquin County, and the broader region, we also encourage readers to consult the Lodi News-Sentinel, Stocktonia, The Sacramento Bee, CalMatters, and other established news organizations. Our full editorial standards and news-media-status statement is published at lodi411.com/editorial-standards.

This LodiEye policy report was produced using artificial intelligence tools under the direction and review of the founder. Lodi411 uses multiple AI platforms in its research and publication workflow, including Anthropic's Claude (primarily Opus and Sonnet models) and Perplexity AI across a variety of large language models offered by each. These tools were used in the following capacities:

Source Discovery: AI-assisted search and retrieval identified roughly two dozen sources, including Pew Research Center survey data, the Belfer Center and MIT security analyses, Congressional Budget Office and Yale labor research, California and EU statutes, the City of Lodi's own records and Legistar meeting archive, and Lodi News-Sentinel and LodiEye reporting. Perplexity AI handled initial source discovery and real-time data retrieval; Claude was used for deeper analysis of the identified sources.

Credibility Validation: AI cross-referenced claims across independent sources, prioritizing government datasets and statutes, peer-reviewed and institutional research, and established news reporting. Multiple models independently verified key figures, and the Yepez “wrong project for Lodi” quote was confirmed against accessible reporting; two further quotes that could not be verified against a primary source were dropped rather than published.

Analysis and Synthesis: Claude Opus and Sonnet helped develop the report's central “steer, don't stop” framework, separating real harms from exaggerated fears, and connected the national argument to the local Lodi case.

Presentation: Claude assisted in drafting, structuring, and formatting the report, including the data-center opinion chart, the comparison tables, and the plain-language pass across the whole report.

Final Review: Multiple AI models reviewed the completed draft for factual consistency, source attribution accuracy, logical coherence, and balanced presentation. Throughout, the editor sets the report's goals, scope, and tone; creates and shapes draft content; reviews and edits the report; integrates independent fact checks; and reviews the AI cross-checks and validations. Multi-tool cross-checking across independent models and sources is the primary error-reduction mechanism.

Lodi411/LodiEye believes transparency about how our research is produced — including our use of AI under human direction — strengthens trust with readers and the broader information ecosystem. Readers who spot an error are encouraged to write editor@lodi411.com so we can correct it.

References

Don Bradford
Next

The Bill That Could Price Lodi Out of Its Own Public Records