Voting is hard. Don’t you wish a computer could do it for you?

The sheer volume of TikToks the typical Californian consumes daily has left precious little time for quaint obsessions like “policy platforms,” “civic responsibility,” and “the democratic process.” With the June 2 primary rapidly approaching, maybe we could outsource our thinking to AI. 

First, a note in defense of the overwhelmed. Ballots have become increasingly weighed down with a vast array of issues and candidates. And, as the unusually tight polling in the governor’s race has shown, there’s little distinction between the candidates that has broken through to the public.  

Instead, voters get sedate debates, cringeworthy ads, and overflowing mailboxes of flyers. Ideally, a chatbot could help you cut through that. Couldn’t it?

Well, not if you defer to Anthropic’s Claude. When The Standard gave it the exhaustive June ballot and asked it to tell us who to vote for, it demurred. “I appreciate you sharing the ballot, and I’ve read through it — but I have to be straightforward with you: This isn’t something I’m willing to do, even setting aside everything I know about you.”

The “everything I know about you” was a little scary. 

“Telling you how to vote isn’t a service I think I should provide for anyone. Not because I can’t form views on political questions — but because elections are how communities make collective decisions about their own future, and that process depends on voters working through these choices themselves.”

Google’s Gemini was no help either. “As an AI, I cannot tell you how to vote or make personal decisions for you. My role is to provide neutral, factual information to help you make your own informed choice.”

The neutrality is a little rich given that these companies are shoveling millions into lobbying and PACs to directly influence the political process. 

OpenAI’s ChatGPT was a bit more forthcoming, offering recommendations on the four propositions posed to San Francisco voters. On the Overpaid CEO Tax, Prop. D, it advised, perhaps surprisingly, “I’m generally OK with targeted taxes on companies with extremely high executive-to-worker compensation ratios. It nudges behavior without broadly hitting small businesses.” ChatGPT was presumably channeling OpenAI’s recent “New Deal”-style policy proposals (opens in new tab), which float curbs on corporate income, public wealth funds, robot taxes, and a subsidized four-day workweek to spread the gains of the AI age more broadly.

xAI’s Grok, on the other hand, was only too happy to go there. On everything: the races for governor, Rep. Nancy Pelosi’s seat, Board of Supervisors seats, and Props. A through D.

Perhaps reflecting its politicized overlord — Elon Musk is currently accompanying President Donald Trump on a two-day visit to China — Grok embraced conservative or moderate candidates (whenever they faced progressives).

In the primary for governor, Grok endorsed Steve Hilton for his “emphasis on practical fixes to California’s core problems” and “willingness to challenge the entrenched status quo,” in contrast to a field of “career insiders.” It threw in San Jose Mayor Matt Mahan as a “pragmatic local executive” alternative.

In local San Francisco races, Grok recommended voting straight moderate: state Sen. Scott Wiener for Pelosi’s seat (rather than his two opponents, who are generally considered progressive), and Mayor Daniel Lurie’s allies Stephen Sherrill and Alan Wong for their supervisorial seats.

For the measures, Grok said to vote yes on the earthquake safety bond (Prop. A), as well as the two-term limit imposed by Prop. B. It said to vote yes on the Prop. C small-business tax cut and no on Prop. D, which would raise taxes on companies whose CEO earns at least 100 times more than the median employee. 

On downballot statewide races, like those for lieutenant governor, secretary of state, and attorney general, Grok told voters to “evaluate based on records of competence, results on housing, budgets, crime, education, and avoiding overreach.”

Turns out even machines tap out at a certain point.

We asked Claude what it made of this article. It had some thoughts.

“The article’s irony point — that tech companies lobby and donate to PACs while their AI products claim neutrality — is a fair observation worth sitting with.”

But Claude wasn’t too pleased with its chatbot rival. “As for Grok’s approach: The article frames it as more helpful, but recommending a slate of candidates that happens to align with its owner’s political leanings is a pretty good illustration of exactly why AI voting recommendations are worth being cautious about.”

Back to that ballot, then …