Washington Should Let America’s AI Flourish

Originally published on Substack. 

The AI fight is hot. Politico reports that the White House is weighing stronger oversight of advanced AI models before release, even as federal officials are already testing some models through national-security reviews.

That tension matters.

Some in Washington still understand that America wins by building, competing, and innovating. Others are drifting toward the same old trap: fear first, bureaucracy second, progress last.

Too many yes-men follow the latest political mood instead of clear principles. Sometimes the White House gets it right, and its AI framework does on one key point: Congress should not create a new federal AI rulemaking body.

That is the right instinct.

AI is too broad, too fast, and too important to put behind a federal permission slip.

AI Is Already Helping People

The strongest case for AI is not theoretical. It is human.

AI is already helping doctors, teachers, workers, students, farmers, banks, and researchers do more with less. NetChoice makes this clear in health care.

One AI-assisted mammography study found 20 percent more cancers detected without increasing false positives. Other tools are improving early Parkinson’s detection. An AI sepsis tool is helping clinicians identify a deadly condition faster.

That matters because delay is not neutral. If AI can help detect cancer sooner, identify sepsis faster, or spot disease earlier, then slowing deployment has a cost. That cost is measured in time, money, and lives.
​
The Pelican Institute adds more examples.

AI-enabled tools are helping researchers detect Parkinson’s and Alzheimer’s earlier and better understand genetic mutations. That is progress. Not hype. Not science fiction. Progress.

The Jobs Panic Is Too Simple

The usual political panic says AI will wipe out work. That is easy to say. It is not what the evidence shows so far.

A recent data review notes that LinkedIn job-posting evidence found 640,000 U.S. jobs created between 2023 and 2025 tied to AI growth. The same review points to survey evidence showing little overall employment effect in 2025, not mass displacement.

That should not surprise us.

Technology changes work. It does not just erase it. Better tools reduce low-value tasks. They increase productivity. They create new services, new firms, and new jobs.

Students are proving the same point. Eighty-seven percent of college students surveyed use AI, often to understand hard material. That is not just automation. That is human enhancement. A serious country should want more people learning faster, not fewer.

AI Helps Locally, Too

AI is not just for Silicon Valley.

The Pelican Institute shows practical uses close to home: banks using AI for fraud detection, local governments improving roads, health systems reducing clinical paperwork, students building apps, and AI reading tools improving literacy.

Texas and other states should see this as a preview. AI can help doctors in Houston, manufacturers in Fort Worth, energy firms in Midland, banks in Tyler, teachers in San Antonio, and small businesses everywhere. This is not about replacing people. It is about equipping them.
​
The AI Opportunity Project by Abundance Institute shows how wide the gains can be: tools that predict 130 diseases from one night of sleep, systems that reduce pesticide use through precision weed control, and applications that personalize athletic training.

Better information means better decisions. Better decisions mean less waste, lower costs, and more opportunity.
​
An FDA for AI Would Backfire

This is where policymakers need radical candor. A broad federal pre-approval regime for AI would be a disaster.

Adam Thierer and Neil Chilson warn against the “false promise of preemptive regulation”.

They are correct. Preemptive regulation focuses on visible risks while hiding invisible costs: startups never launched, tools never deployed, patients never helped, and productivity never gained.

Joe Lonsdale made a similar point in his CNBC interview. Any national review for AI should be “as limited and targeted as possible.”

He also warned that a sprawling AI bureaucracy would slow innovation, entrench big firms, and give China room to gain ground.
​
That is the danger.

Big companies can afford lawyers, lobbyists, compliance teams, and delays. Startups cannot. A complicated federal review system would not just “protect consumers.” It would protect incumbents.

That is how regulatory capture works.

Good Rules Are Narrow

None of this means no rules.

Fraud should be punished. Theft should be punished. Cyber abuse should be punished. Real national-security threats should be addressed. But that is very different from building a broad federal gatekeeping regime for a general-purpose technology.

The better path is clear: use existing law where possible, fill real gaps carefully, and keep guardrails narrow, fast, and tied to actual harms. Pelican’s AI Toolkit gets this balance right by focusing on practical solutions that address concerns while maximizing benefits.

Markets are discovery engines. Bureaucracies are not.

AI will improve fastest when millions of people can test it, use it, reject bad tools, and scale good ones. No federal office can predict every use in medicine, education, energy, finance, agriculture, and logistics.

Three Key Takeaways for Policymakers

• AI is already delivering real benefits. It is improving cancer detection, sepsis response, fraud prevention, literacy, infrastructure management, agriculture, and scientific discovery.
• A broad pre-approval regime would punish challengers. It would raise costs, protect incumbents, and slow useful tools before they reach the public.
• Good AI policy should be narrow and harm-based. Police fraud, theft, cyber abuse, and national-security risks, but do not smother innovation with a federal permission slip.

America should choose confidence over panic.

Tell lawmakers to reject a broad AI bureaucracy. Keep guardrails narrow. Defend the freedom to build. Let markets work so AI can help people prosper.