Deconstructing the U.S. approach to AI: What’s being sacrificed at the altar of unfettered innovation?
Unsplash
Unsplash· 6 min read
The Trump Administration’s signing of the Ensuring a National Policy Framework for Artificial Intelligence Executive Order (AI EO) last month sounds lofty and promising, but the devil is in the details. It is less about creating a federal framework for AI and more about dismantling state–level enacted stakeholder AI protections. It is less about the unfurling of unfettered innovation and the protection of US national security and more about a fast and furious lining of the pockets of the US based billionaire technology class.
The AI EO is not good for stakeholder care. It is not good for sustainable innovation. It is not good for national security. And it is not good for the rule of law. Indeed, each of these important objectives is being sacrificed at the altar of unfettered innovation, benefiting the very few.
Let’s start with stakeholder care. Far from being a national framework that one would assume protects the nation and its inhabitants, it is aimed at eliminating State AI laws created to protect stakeholders from the downsides of AI - because nobody else is doing it (here’s looking at you, US Congress). And it is about anointing the very few – Silicon Valley billionaires and companies ideologically aligned with Trump - with the unbridled, unfettered pursuit of innovation profit.
The AI EO unfortunately extends the political polarization that has been growing in Silicon Valley over the past few years between the so called “tech optimists” or “accelerationists” (Andreesen, Thiel, Musk, Zukerberg) or what I call the “Tech Masters of the Universe” in my forthcoming book, Governing Pandora, and what the accelerationists call “doomers” or “decelerationists” who they claim are luddites. Instead, I call these folks the “Tech Guardians of the Universe” because they advocate for the privacy, human, and democratic rights of affected populations while still advocating for tech progress. Tristan Harris of the Centre for Humane Technology, Max Tegmark of the Future of Life Institute, and even the Godfather of AI, Geoffrey Hinton, come to mind for this category.
It is no coincidence or happenstance that the so-called White House AI and Crypto Tsar, David Sacks, who arguably has the most influence on tech policy in the US right now, is a Silicon Valley tech billionaire. Many times over. And a proud accelerationist. And continues to be invested in all manner of AI and tech interests while holding the top tech role in the Administration.
Don’t get me wrong – I am all for unleashing tech infused innovation as it has vast potential to discover unheard of benefits for humankind in health, education, business, science, and so much more. But such progress must happen with appropriate guardrails to protect against the material, severe (and maybe even existential) downsides that we know about and know nothing about.
Here’s a constantly updated repository of AI risks maintained by MIT to give you a sense of the range of issues we know something about and those we don’t. Here’s the EU AI Act, a pioneering law that classifies AI risk into four categories (Unacceptable-, High-, Limited-, and Minimal-Risk AI Systems) and provides guidance on how to deal with each.
There is nothing like this in the US (although the Biden Administration issued its own EO addressing AI risks, superseded by the current AI EO). Indeed, the current administration either insults the EU over its focus on stakeholder care or threatens to retaliate against it for various reasons, including the regulation and fining of US tech companies that haven’t complied with EU laws.
Now, the US administration seems to be applying a similar approach within the US against States (California, New York, Colorado) bold enough to pass their own AI laws in the absence of federal laws. The AI EO contains a variety of aggressive policy positions (which may or may not withstand legal challenge from States and other affected stakeholders with standing). Among them is a direction to the US Attorney General to create a task force to challenge State laws and the Commerce Department to draw up a list of problematic state rules. The AI EO also threatens to withhold broadband programs and funding from States with AI laws.
Pinch me, but I don’t recall this sort of brazen federal/state policy bullying in the decades I’ve been alive. So instead of a national framework and common safeguards, we are getting the worst of both worlds – the targeted elimination of guardrails that exist in some key States, and no federal replacement for such guardrails.
Instead, the purpose of the AI EO seems strictly focused on unleashing the animal spirits of innovation, the speed and scale of which we have never seen before. Because before we never had technologies – except for nuclear technology - that we didn’t fully understand, with the potential of wreaking widespread harm and even destruction. Even in the case of nuclear tech, it was in the hands of the very few, under relatively tight and effective controls, the threat of mutual assured destruction, and was extraordinarily difficult to replicate. In contrast, AI is everywhere all at once, usable by anyone, potentially autonomous, and geometrically or infinitely replicable.
Then there’s the issue of “national security”. It is most definitely a critically important part of the equation. Indeed, most Americans would agree that we would prefer to be the dominant global AI force as opposed to seeing China “win” that race. However, it is arguable that the present approach that benefits the interests of a tiny minority of wealthy individuals and companies laced with conflicts of interest, cross-investing, government favoritism (including who can sell what to whom, even China), and what appears to be, to the trained eye, bribery and corruption, is deeply dangerous to national security.
While the AI EO may sound good, it: reduces stakeholder protections by targeting the elimination of State laws protecting stakeholders; threatens the safe and sustainable development of innovation by eliminating guardrails potentially leading to something going really wrong; and undermines US national security and the rule of law by advocating wild and unregulated growth benefiting the very few who may have attained their stature through dubious legal means.
What to do? In the US, work hard on protecting States’ rights or enact a comprehensive federal AI law. Internationally, collaborate instead of bully other jurisdictions like the EU that have deep expertise and thoughtful approaches to AI. Overall, incentivize responsible and sustainable innovation, protect national security by enforcing laws and holding allies near, and return to effective rule of law where ethics, integrity, accountability, and transparency actually mean something.
illuminem Voices is a democratic space presenting the thoughts and opinions of leading Sustainability & Energy writers, their opinions do not necessarily represent those of illuminem.
See how the companies in your sector perform on sustainability. On illuminem’s Data Hub™, access emissions data, ESG performance, and climate commitments for thousands of industrial players across the globe.
illuminem briefings

AI · Corporate Governance
Andrea Bonime-Blanc

Green Tech · AI
illuminem briefings

Power Grid · AI
The Wall Street Journal

AI · Ethical Governance
The Washington Post

AI · Public Governance
energynews

Green Tech · Sustainable Business