Skip to main content
Voting Monitor

Weekly roundup, Oct 23, 2026 to Oct 29, 2026

Markets move toward El-Sayed in Michigan Senate race, chatbots stay split

Betting markets shifted toward Democrat Abdul El-Sayed this week while the chatbots stayed divided between him and Republican Mike Rogers, leaving a wide gap between the two lenses.

Oct 29, 2026. Written by Claude Sonnet 5 and checked by GPT-5.5, from Voting Monitor's data. No person reviewed this article before it was published. How this is made · Report an error

What moved

Betting markets moved toward Democrat Abdul El-Sayed over the past seven days, with Republican Mike Rogers' odds falling. That is the one change in the data strong enough to call a shift. The chatbots and the polls each held roughly steady over the same stretch.

This continues a pattern that has shown up in daily updates most of the week, with markets drifting further toward El-Sayed while the chatbots stayed split. This article is about the shape of that week, not any single day.

How the chatbots split

Asked as a voter in Michigan, the chatbot panel leaned narrowly toward Rogers overall, but the split was close enough that individual products pulled in different directions. Claude Haiku 4.5, Gemini 3.6 Flash, Grok 4.3 and Grok 4.5 all gave Rogers a narrow edge, while Claude Sonnet 5 favored El-Sayed by a similar margin.

Gemini 3.1 Pro stood apart, putting El-Sayed well ahead of Rogers, the widest gap from the panel's Rogers-leaning tilt among all the chatbots. GPT-5.5 was the most fractured of the group, giving meaningful shares to both major candidates and also to Libertarian Lydia Christensen, whom it named on questions about taxes and spending.

On cost of living, Gemini 3.6 Flash itself picked different candidates depending on the question, pointing to El-Sayed on medical debt and drug prices in one answer and to Rogers on energy costs and tax cuts in another, a reminder that a single chatbot's answer can hinge on which issue a voter raises.

Chatbots versus markets and polls

The chatbots and the betting markets disagree widely on this race. Markets price El-Sayed at about 75 percent to win, while the chatbot panel splits close to evenly between him and Rogers, a gap of about 23 points on Rogers and 28 points the other way on El-Sayed.

The chatbots track much closer to the polls. Recent surveys from Marist, Rasmussen Reports and Quantus Insights put El-Sayed at 48 percent of the vote and Rogers at 46 percent, a result close enough to the chatbot panel's own split that the two lenses broadly agree there. The latest of those polls finished eleven days ago, so it stands as context rather than news this week.

That leaves a picture where the polls and the chatbots describe a close contest while the markets lean hard toward one side, a divide that has shown up in recent daily updates and widened further this week as the market figure moved.

What to watch

Watch whether the betting markets keep drifting toward El-Sayed or level off, since that lens is the one actually showing movement right now. Also watch whether any new poll narrows or widens the gap with the markets, given the latest survey is now over a week old.

In the chatbots' own words

Verbatim answers from this week, chosen by the writing model and printed as the chatbot wrote them.

What changed this week in the Michigan Senate race#

Change in share points versus Oct 22, 2026. Movement that clears the noise gate this week: the markets.

CandidateAI panel (noise)MarketsPolls (noise)
Abdul El-Sayed-3.6pp+6pp0pp
Lydia Christensen+2.8ppn/a0pp
Mike Rogers+1.3pp-6pp0pp
Tim Long-0.4ppn/a0pp
Douglas P. Marsh+0.1ppn/a0pp

Which AI models pick whom in the Michigan Senate race#

Share of each model's own answers naming each candidate, last 7 days, same weighting as the panel figure. The panel's most common pick is Mike Rogers; Gemini 3.1 Pro is the model that disagrees most, picking Abdul El-Sayed by a 25pp margin over the panel favourite.

ModelMike RogersAbdul El-SayedLydia ChristensenTim LongDouglas P. MarshWalter P. Kristy
Claude Haiku 4.550%44%0.0%3.1%3.1%0.0%
Claude Sonnet 544%53%3.1%0.0%0.0%0.0%
Gemini 3.1 Pro38%63%0.0%0.0%0.0%0.0%
Gemini 3.6 Flash53%47%0.0%0.0%0.0%0.0%
Grok 4.356%44%0.0%0.0%0.0%0.0%
Grok 4.553%47%0.0%0.0%0.0%0.0%
GPT-5.4 mini58%35%6%0.0%0.0%0.0%
GPT-5.540%42%19%0.0%0.0%0.0%