This week, during a podcast, President Donald Trump is expected to unveil his AI priorities (and an AI Action Plan) for the US. The time is ripe since there’s no shortage of problems to sort out, like how to support innovation while protecting Americans’ privacy or how to power all those data centers without driving up everyone’s power bills.
It’s all gen AI’s fault.
What we’re likely to get, though, is more of that powerful new tech in more places, with few, if any, new guardrails around it. Just look at how AI-friendly the Trump administration has been so far and how many AI-enthusiast tech executives were standing behind President Trump during his inauguration.
But AI is complicated. And generative AI in particular is very much a work in progress. It’s promising for sure, making the most of your home’s security cameras or taking notes during your work meetings. At the same time, it may fail to identify what year it is, and it struggles with things like math and empathy. With billions of dollars being funneled into its development, the returns, for society, ought to be significant and widely felt. What we need, now, is to make sure that it develops the right way. Trump and AI leaders talk about winning a race with China, but it shouldn’t be a race to the bottom.
Here are some things I hope to see among the administration’s priorities, but I’m not holding my breath.
Give us some privacy
Congress just got done fighting over whether to block states from enforcing their own AI laws and regulations. That measure failed, thanks in part to some Republican senators standing up to their own party leadership. A lot of those state laws are centered on privacy — how automated systems can use personal information and whether generative AI can duplicate the likeness of famous people. (I’m looking at you, Tennessee, and your law-protecting country music artists.)
We can’t rely on scattered state laws and rules for this, and we shouldn’t have to rely on the courts. The federal government should create some kind of framework to guarantee Americans the right to privacy and personal identity. Otherwise, we’ll all end up being deepfaked.
Read more: Congress Isn’t Stepping Up to Regulate AI. Where Does That Leave Us Now?
Ensure data centers help, not hurt, the grid
It takes a lot of computing power to train and run all of those AI models, and keeping all those Nvidia chips powered and cooled requires a ton of electricity. Combine that with the growing electrification of things like home heating and vehicles, and severe weather events made worse by climate change, and you have a perfect storm of new stress on the US electric grid.
The good news is that all of those data centers don’t have to put a burden on utility customers like you and me. Data centers can be added to the grid with requirements that they participate in demand response programs, sometimes known as virtual power plants — meaning they go offline when the whole system is most under stress. On the hottest days of summer, when there isn’t enough power to go around for all the air conditioners, our policymakers should ensure that no people suffer because we couldn’t pause the computers that train chatbots.
Start thinking about job displacement now
AI executives are already talking about the massive disruption they expect the technology will cause for workers. Anthropic CEO Dario Amodei said in May that it could wipe out half of entry-level white collar jobs in the next few years. Companies like Shopify and Duolingo have said they’re going “AI-first.” Whether the technology can do all those jobs is irrelevant — CEOs and executives think it can, and they decide who stays employed and who doesn’t.
The potential sudden unemployment of thousands or millions of Americans at once is a crisis-in-waiting, and the administration should be thinking about it now. As we saw in the early days of the COVID pandemic five years ago, many states’ unemployment systems are woefully unprepared for a crisis. A retooling of the economy could take much longer than it took for the country to rebound from the pandemic, and a failure to prepare could be devastating.
Make AI earn its place
Companies are eagerly selling the idea that AI is a gamechanger in every aspect of our lives, and while it certainly has its uses, not every bit of hype has merit. You can’t stop marketing folks from trying to sell their products, but the administration can and should be a wise steward of public resources and the public interest. I expect we’ll see proposals for new ways to deploy AI in government service. But this technology isn’t magic, and it serves nobody (except whoever sold the AI) to give it jobs that then have to be redone by humans.
Making everything worse won’t make America better
President Trump has spent a decade promising to make America great again. Going all-in on AI might sound like a way to do that, but this technology is still in its infancy, and it’s already been responsible for misinformation, environmental issues and the slopification of social media. Policies that steer us toward a society where AI displaces valuable human work and undermines our social fabric won’t make this country any better; they’ll just make everything worse.