It’s poetic that the same day Darline Graham won her primary to seek a full term in the United States Senate, the Wall Street Journal admitted it had published an AI-written op-ed and, based on opinion page editor Paul Gigot’s remarks about it, would happily do so again. “AI is a fact of modern life. People will use it to assist in their work and their writing, including with research, checking grammar, editing and more.”
On social media, journalists at the New York Times and elsewhere were quick to voice objections to Gigot’s leniency, but others, including one at the Washington Post, shot back at the critics: how was publishing an AI-written op-ed any different from publishing a ghost-written one?
She’s gaffe-prone. If that’s a symptom of humanity, she should not be quick to seek a cure
Rich people, politicians, Hollywood actresses and tenured academics with enough clout to get a trade-publisher contract frequently don’t write their own books or op-eds. Elected members of Congress hardly even write their own legislation. That’s what staff or lobbyists are for.
So it’s hard to say Darline Graham isn’t qualified to be a US senator, whether or not she knows what DACA is. (For those who don’t want to ask the AI that’s taken the place of Google’s search engine: it’s “Deferred Action for Childhood Arrivals,” a scheme that allows illegal immigrants who came to the country as minors to remain here.) Senator Graham II inherits her late brother’s staff and interest-group friends: his “office,” which matters more than the elected official these days, as the examples of more-dead-than-alive senators like (the now late) Dianne Feinstein and (only getting later) Mitch McConnell prove.
We reached the point earlier this year where a sitting senator had to be propped up and posed with the day’s newspaper as proof of life. Senator Graham II at least makes a steady blip on the electroencephalogram. That can’t be taken for granted in the “world’s greatest deliberative body”–emphasis on body. Its members’ bodies deliberate every day on whether to shut down or not.
Where mind and soul are concerned, there’s not much to deliberate about in 2026, unless it’s over the question of whether the AI that are replacing us have minds and souls. Science has already decided that we don’t, and our robotic behavior supports the conclusion. Human beings have built artificial intelligence to do what we can’t: think, even feel. And write op-eds and laws.
The irony is that Graham’s nomination to stand for a full term is as much a reaction against political autopilot as an affirmation of it. Ralph Norman, the man she beat in the runoff primary, has a rock-solid conservative record laid down over nearly a decade’s service in the House of Representatives. If Grok had been asked to choose a candidate on their merits, it or any other reasonable AI would have chosen Norman. Principled voters who apply machine-like consistency to their decision-making would have selected the logical choice, too, following standard criteria of ideology and objective qualifications. Electing Graham, with her demonstrated ignorance of issues over which she could wind up having a decisive say, was in defiance of sound method as well as the demands of principle. You’re supposed to listen to the experts, who tell you which candidate best matches coherent ideological metrics, then you punch the right ticket. You don’t even have to make a choice, really: it’s made for you by your principles. All you have to do, or all someone you trust has to do for you, is assign objective grades to the candidates.
Darline Graham earned herself a failing grade by just about any metric in her one debate with Norman. Yet South Carolina Republicans opted for her anyway, out of human reasons rather than mechanically logical ones. She was a beneficiary of loyalty: personal loyalty to her late brother and GOP voter loyalty to her political sponsor, President Trump. She was a beneficiary of sympathy, too, over her brother’s death. And while Norman may have had better credentials as a populist in principle, Graham had the populist benefit of a lack of experience. She went into the runoff with two of the advantages Donald Trump had in the 2016 Republican presidential primaries: she was clearly not a member of the Washington political class, however beholden to it she may be primed to become, and she enjoyed maximal name-recognition. That turned out to be a winning formula against someone whose conservative bona fides were infinitely better established. Trump easily beat all the 110 percent Reagan conservatives he faced in 2016. Whatever the merits of Reagan conservatism or principled conservatism of any kind, a presumptive claim upon the hearts of Republican voters is not among them. They don’t vote like machines; they vote like family.
That can be a good thing or a bad thing; either way, it is, as Paul Gigot says about AI, “a fact of modern life.” Voters don’t value experience or ideological coherence nearly as much as they value a feeling of kinship. This infuriates rationalists and systematizers of every variety, including those who want to systematize populism. But elections aren’t about systems and how beautifully coherent they are; they’re about voters.
The work of governing, however, tends to turn in the opposite direction, toward the demands of experts, narrow interests, ideologues, insiders and people proud to shun disreputable relatives. Ghost writers once served as a link between the elite and popular levels of government. They could make a dull-witted politician sound articulate enough for Washington; or they could make a sophisticated official sound inspiring to ordinary people. Now AI can provide these social translations – if the words, or even the thoughts, are not the politician’s own, that was true before AI as well. Government was becoming steadily more mechanical, more professionalized and specialized, long before the advent of AI.
It was also becoming steadily less legitimate in the eyes of voters. Gigot’s defenders are right that ghost writers were doing what AI can now do. But look at the results those ghosts were producing – epidemic levels of cynicism and popular dismissal of politicians’ words. Most Americans may not know how thoroughly scripted their elected leaders are – that they don’t write their own books, op-eds or speeches. Yet they feel what they don’t know. They sense the phoniness of it all. And that’s when human beings are doing the ghosting. What will machine-scripted politicians – and elites of all kinds – reap?
Darline Graham went off-script several times during her brief campaign for the nomination. She’s a curious hybrid of insider and outsider, on account of her inexperience if nothing else. She’s gaffe-prone. If that’s a symptom of humanity, she should not be quick to seek a cure. She should defy programming and Washington’s programmers. That probably won’t happen, but people are unpredictable, and sometimes for the better.
On the flipside, populists who want to systematize their ideas and translate them into effective policy should be careful not to forget where the demand for populism is coming from. Americans want self-government, not government by ghosts, machines and the Senate’s living dead.
Comments