9 min read

ChatGPT is a Romney-Biden voter

Alas, voters aren’t robots.
ChatGPT is a Romney-Biden voter
Illustration by DALL-E

Good morning! It’s Tuesday, June 11, 2024. Election Day is 147 days away. If this newsletter was forwarded to you, subscribe here. If you want to contribute to support my work, donate here.

Breaking: Hunter Biden has been found guilty of three felony counts of lying on a federal firearm form and illegally owning a gun. He is the first child of a president in U.S. history to be a convicted felon. I’ll have more analysis on the conviction in tomorrow’s newsletter. Now, for something completely different:

ChatGPT doesn’t like talking about politics.

Ask it for an opinion on pretty much any politically charged matter, and the chatbot will beg off. Ask it who you should vote for in the 2024 election, and you’ll get a similarly cagey response. “As an AI language model, I don’t have personal opinions or preferences, so I can’t tell you who to vote for,” the OpenAI bot told me when I asked it to pick between Joe Biden and Donald Trump. “Your decision should be based on your own values, beliefs, and priorities.”

However, given the right prompting, the chatbot is all too happy to engage in hypotheticals. I recently told it that — like a plurality of Americans — my top issue was the economy. I mentioned four economic indicators I was paying attention to (inflation, unemployment, GDP growth, and income levels), and asked it for benchmarks for when I should consider voting for or against an incumbent president’s re-election. (For each one, it gave answers that were in line with most economists.)

Then, I gave it various sets of economic data and asked for its advice. Given the numbers from 2012 and 2020, ChatGPT recommended voting against the president’s re-election. Given the numbers from 2024, it said to give the incumbent another term. (Notably it also made that recommendation with the numbers from 2019, the last pre-Covid year of Donald Trump’s presidency.) ChatGPT acknowledged that “inflation may be a concern” in the 2024 scenario, but it was impressed enough with the unemployment and GDP figures to recommend staying the course. “The overall economic indicators present a reasonably healthy economy,” it said.

Of course, in the real world, the president did lose in 2020 — but incumbent Barack Obama won in 2012 with 332 electoral votes, despite what ChatGPT judged to be a “mixed economic performance.” Meanwhile, in 2024, Joe Biden will be lucky to scrape by with 270 electoral votes, if he even wins. Democrats remain enduringly puzzled by this disconnect.

However, it’s not as hard to understand once you remember that voters, unlike ChatGPT, aren’t robots. You can’t feed economic data into them and expect them to all spit out the same result. As ChatGPT itself told me when I asked about this, “While economic indicators such as low unemployment and GDP growth may be positive, if people don’t feel the benefits in their daily lives, they may still be dissatisfied.” It also noted that “the incumbent’s messaging and communication strategy may fail to effectively convey their achievements or may be overshadowed by other news events or controversies.” If the president has a low enough approval rating, and the challenger is able to skillfully paint a “narrative of economic mismanagement,” the incumbent could still face problems.

Conversely, if the incumbent is a talented messenger or if he’s able to “shift blame for economic challenges onto external factors or previous administrations” — a la Obama 2012 — they might be able to outperform subpar economic numbers. Elections come down to more than just objective data.

A global blind spot

This is hardly a new phenomenon. In his 2022 book, “Blind Spot,” Jon Clifton — who spends his days steeped in more subjective data, as the CEO of the polling firm Gallup — collects examples of times when populations demanded political change despite economies that seemed perfectly rosy on the surface.

Here, for example, are charts showing GDP per capita in Egypt and Tunisia in the two decades before 2010. You wouldn’t exactly look at these trajectories and expect the Arab Spring.

Graphs from “Blind Spot” by Jon Clifton

Clifton, however, believes he’s found a measure — albeit a more subjective one — that was able to accurately predict the uprisings in these economically well-performing countries: happiness. In the years leading up to the Arab Spring, Gallup asked Egyptians and Tunisians, “How is your life going?” Even as the economy grew and grew, the polling firm found that the percentages of people in those countries who rated their lives high enough for Gallup to classify them as “thriving” was dropping precipitously. Then, the protests began.

Clifton shows a similar pattern in the United Kingdom before Brexit and in the United States before the 2016 election: objective economic indicators suggesting that everything was A-OK, while measures of happiness flared flashing red alarms. He also presents research from MIT’s George Ward, who studied European elections and found that measures of happiness were the best at predicting whether voters kept a sitting government in power, beating out the predictive power of GDP growth, unemployment, or inflation.

“Imagine if these were unemployment trends,” writes Clifton. “A 10-15-point swing in unemployment anywhere would be headline news everywhere. Tunisia’s 10-point drop [in happiness] meant that 800,000 adults who were thriving in 2008 were no longer thriving in 2010. The 15-point drop in Egypt meant that almost 9 million people who were thriving in 2006 were either struggling or suffering in 2008.” But because the drops took place away from the types of indicators that global leader usually pay attention to, they went completely under the radar.

The incumbency disadvantage

These numbers on what leads countries to ditch their incumbent governments seem newly relevant at a moment when a lot of countries are choosing to ditch their incumbent governments.

That was one takeaway from the European Union elections last weekend, in which the incumbent leaders of many countries — including France’s Emmanuel Macron, Germany’s Olaf Scholz, and Hungary’s Viktor Orbán — saw their parties struggle at the polls. Similarly, incumbent Rishi Sunak is cruising for a bruising in the UK’s elections next month, while Canada’s Justin Trudeau is deeply unpopular and India’s Narendra Modi lost ground in his country’s recent elections. And then, of course, there is Joe Biden, who certainly could still win re-election — but is also mired in historically low approval ratings, as voters consistently express doubts about his economic performance, foreign policy, and mental acuity.

To explain all this, we know where Clifton would point: Gallup’s latest numbers on the global mood, which reveal a world population that has grown steadily less happy over the last few years. Here in the U.S., we fell out of the top 20 happiest countries for the first time last year. Gallup attributed much of that slide to a sharp drop in happiness among young people — who are, as I’ve noted, one of the groups most averse to keeping Joe Biden in power in 2024.

Clifton notes that the global drop in happiness didn’t start with Covid — but it isn’t hard to understand how the pandemic might have supercharged it. In his book, Clifton writes that Gallup’s surveys have yielded five indicators that are shared by people who are happy with their lives: “They are fulfilled by their work, have little financial stress, live in great communities, have good physical health, and have loved ones they can turn to for help.”

That is also a list of five indicators that, for most people, would have suffered greatly during Covid.

When I think about voters’ dissatisfaction with incumbent administration post-Covid, my mind naturally wanders to the administration that I most closely lived under these last four years: the one at university. My time at Georgetown, where I matriculated in 2020 and graduated in 2024, doesn’t map perfectly onto Joe Biden’s time in office — but it comes pretty close.

By the time I graduated, there were no visible signs of Covid left on campus: the mask and hand sanitizer stations were gone, the “six feet apart” signs had been taken down. Day to day, I doubt most students thought of the pandemic. And yet, when you talk to most students in my graduating class, it doesn’t take long to pick up on an underlying sense that things are different than they were before. People will usually mention small things, but they add up to a fairly widespread confidence that Covid changed our lives on campus. The dining hall isn’t open as late as it used to be. Bars that upperclassmen used to frequent have closed their doors. Involvement with clubs and relationships with old professors, both of which had to be forged behind masks, are somewhat dimmed. And then, on top of all that, there’s a leftover frustration with the fact that we were kept off-campus for our entire freshman year, denying us a year of college life. (There’s a reason Axios recently ran article with the headline, “How the Class of 2024 lost college.”)

My guess is most of you can relate in your own ways, whether it’s a relationship that’s changed since Covid or a workplace that has never quite gone back to normal. It may not be the first thing on your mind every day, but it might contribute to a lingering dissatisfaction that sits there nonetheless. I think Joe Biden is suffering from this same problem.

Certainly, if you ask most Americans, they can relate. According to Gallup, most of the country says the pandemic is over — but 57% say their lives have still not gone back to normal. 43% say they don’t think their lives ever will. That probably means different things to different people. Maybe they’re referring to the rise of work-from-home, or to their child’s lasting learning loss. Or, perhaps, they’re thinking more broadly. Biden premised his 2020 campaign on a Warren Harding-esque return to normalcy, and yet, the world is experiencing more armed conflict than at any point since World War II, border crossings have surged to a record high, and U.S. mental health (especially among young people) has dipped to an all-time low.

Some of these problems are outside of Biden’s control, of course — but that’s not exactly an ironclad political defense. Voters generally don’t like to hear that the American commander-in-chief isn’t in control.

Economic data and other traditional indicators of a country’s well-being are important, but they can only get us so far. When trying to understand Joe Biden’s polling — and the electoral losses experienced by other global leaders — the more subjective, but enduring, mood swing since Covid should probably be taken into account.

As Clifton notes in “Blind Spot,” economists have spent the last few decades grappling with the reality that people don’t always act like perfectly rational robots, leading to the rise of behavioral economics. Political operatives should remember that same lesson. A lot goes into someone’s decision at a voting booth: economic numbers, yes, but also their emotions, their happiness, what their friends have told them, what they saw on the news, how they feel that things are going. The tiniest things can change our sentiments: in one humorous example, researchers have found that the smell in a room can impact the political opinions respondents express in a survey.

If something as slight as smell can unconsciously change our attitudes, it shouldn’t be surprising that a mental-health-wrecking pandemic may have done so. My new political adviser, ChatGPT, agrees. “Pandemics can be transformative events that reshape political landscapes, ideologies, and voter attitudes in both the short and long term,” it tells me.

At a time when incumbent leaders are struggling worldwide, perhaps their campaigns should be spending as much time dissecting measures of happiness as they do the latest economic data.


Thanks for reading.

I get up each morning to write Wake Up To Politics because I’m committed to offering an independent and reliable news source that helps you navigate our political system and understand what’s going on in government.

The newsletter is completely free and ad-free — but if you appreciate the work that goes into it, here’s how you can help:

If you have any questions or feedback, feel free to email me: my inbox is always open.‌‌‌‌

Thanks so much for waking up to politics! Have a great day.‌‌‌‌

— Gabe