Vibe Governing: A Cautionary Tale
On April 2, 2025, President Trump announced reciprocal trade tariffs. Within days, the US stock market plunged nearly 20%, wiping out over $6 trillion in wealth. Retaliatory tariffs from global partners swiftly followed, igniting a global trade war.
But how did we get here—and what does AI have to do with it?
Let’s rewind to 2011. Peter Navarro, now President Trump’s Senior Counselor for Trade and Manufacturing, co-authored "Death by China," advocating aggressive tariffs to counter what he perceived as China's unfair trade practices. Fast-forward to April 2025, Navarro’s theories have transformed into policy.
Following the tariff announcement, observers quickly noted something striking: the formula used by the administration mirrored what any AI could effortlessly generate. Financial journalist James Surowiecki was among the first to highlight that reciprocal tariffs appeared calculated simply by dividing the US trade deficit with a country by that country's exports to the US—an approach easily replicable by prompting an AI model to suggest a "fair" tariff system.
Indeed, when I tested this, the AI delivered identical results, including disclaimers such as, "this assumes no retaliation or market disruption" and reminders that such simplistic calculations "don't account for services or complex global supply chains." It even proactively suggested follow-up tasks, such as building a Python calculator or detailing specific country tariffs. Astonishingly helpful—and chillingly simplistic.
Did the Trump administration use AI for their tariff policy? Commerce Secretary Howard Lutnick denied it in a CBS News interview. Nevertheless, one wonders why certain remote areas such as Heard and McDonald islands (which have no permanent human population) were included on the tariff list of 60 countries.
Was this a computer output that no one checked?
The Trump administration has used AI tools in the past. NBC News reported that the government employed AI to analyze responses from federal employees about their weekly accomplishments. The AI assessed whether their roles were essential, aiming to cut the workforce. AXIOS reports that the US State Department uses AI to scan social media activity of student visa holders for “pro-Hamas” or “antisemitic” content. Visas revocations are automated under this program. These sorts of reviews have been done in the past but were done manually. The employment of AI just gives them that much more reach and speed.
The possibility that AI could generate these tariffs, even if it didn't happen this time, demonstrates a potential negative impact of AI unlike scenarios typically imagined, like bioterror or cyberwarfare. It highlights the risk of relying on AI for complex decisions based on simplistic inputs or "vibes" rather than nuanced understanding. AI is a tool; it must be used wisely. This cautionary tale suggests two key approaches:
Always keep a human critically engaged. This means more than just being "in the loop"; it requires active oversight at every stage. Humans must critically evaluate the prompts given to AI, question the data sources it might use, validate outputs against real-world knowledge and context, probe the answers, understand the stated and unstated limitations, and check for nonsensical results (like tariffs on uninhabited islands).
Curate data rigorously. Especially for key policy or business roles, ensure the data informing AI is of the highest quality. AI amplifies the old adage “Garbage in, Garbage out”. The internet contains good information, but also speculation, opinion, and misinformation. An AI researching trade deficits might incorporate Navarro’s 2011 opinion book simply because it's accessible online data.
In 1971, cartoonist Walt Kelly wrote a Pogo comic strip where he says, “We have met the enemy, and he is us”. The line has since become a widely quoted expression of self-awareness and accountability. As we integrate AI into critical functions, this cautionary tale highlights the paramount importance of human oversight, critical thinking, and ultimate accountability in governance. We must learn to use AI wisely.
