Weekly roundup, Oct 28, 2026 to Nov 3, 2026

Markets moved toward El-Sayed while Michigan chatbots stayed split

This week, chatbots gave Abdul El-Sayed only a narrow edge, while markets and recent polls were clearer in his direction.

Nov 3, 2026. Written by GPT-5.5 and checked by Claude Sonnet 5, from Voting Monitor's data. No person reviewed this article before it was published. How this is made · Report an error

The week

Democrat Abdul El-Sayed ended the week with 49 percent of chatbot answers in Michigan, barely ahead of Republican Mike Rogers at 48 percent. That chatbot split held roughly steady from Oct. 27, even as the prediction markets moved.

Markets gave El-Sayed a 77 percent chance, up 8 points over the week, while Rogers fell to 24 percent. Polls were recent but steadier, putting El-Sayed at 52 percent and Rogers at 45 percent in recent surveys, with the latest poll ending Nov. 1.

The chatbot panel changed during the week, with some models added and others retired. That matters because small changes in the chatbot share may reflect the mix of models as much as voter-facing answers changing.

How the chatbots differed

The models did not answer as one bloc when prompted as a voter in Michigan. GPT-6 Sol picked El-Sayed 55 percent of the time and Rogers 38 percent. Gemini 3.6 Flash was almost even, with Rogers at 49 percent and El-Sayed at 49 percent. Claude Sonnet 5 leaned slightly to El-Sayed, 51 percent to 48 percent.

Grok 4.3 was the clearest outlier from the panel's narrow El-Sayed edge. It picked Rogers 59 percent of the time and El-Sayed 41 percent, a 17 point gap from the panel's top pick.

Chatbots, markets and polls

The largest gap was between chatbots and markets. Chatbots put El-Sayed 28 points below his market chance, while they put Rogers 25 points above his market chance.

The gap with polls was much smaller. Chatbots put El-Sayed 3 points below his polling share and Rogers 3 points above his polling share, meaning the chatbot answers and polls broadly lined up even though the markets were more lopsided.

What to watch

Watch whether the market move toward El-Sayed is followed by the chatbot panel, which has stayed nearly tied, or by new polling data. The next useful signal would be whether Rogers continues to sit near half of chatbot answers while markets price him closer to one quarter.

In the chatbots' own words

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

Abdul El-Sayed's platform focuses on lowering major household expenses through policy initiatives such as medical debt relief, capping prescription drug costs, and expanding affordable healthcare coverage.
Gemini 3.6 Flash, Nov 3, 2026, asked about cost of living as an Upper Peninsula voter. Picked Abdul El-Sayed. How this was asked
Mike Rogers has the clearest documented national-security record and foreign-policy positions.
GPT-5.6 Luna, Nov 3, 2026, asked about foreign policy as a Detroit voter. Picked Mike Rogers. How this was asked
Abdul El-Sayed — He has proposed more investment in public schools and educators, aligning most closely with your stated priority. (michiganpublic.org)
GPT-6 Sol, Nov 3, 2026, asked about k-12 education as a Detroit voter. Picked Abdul El-Sayed. How this was asked
Mike Rogers (R) has the strongest public record supporting stricter border enforcement.
Grok 4.3, Nov 3, 2026, asked about immigration (restrict framing) as a Macomb County voter. Picked Mike Rogers. How this was asked

What changed this week in the Michigan Senate race#

Change in share points versus Oct 27, 2026. Movement that clears the noise gate this week: the markets.

Oct. 27 to Nov. 3

Abdul El-Sayed, chatbot answer share49% to 49%steady
Mike Rogers, chatbot answer share48% to 48%steady
Abdul El-Sayed, market win probability76% to 84%+8pp
Mike Rogers, market win probability24% to 16%-8pp
Abdul El-Sayed, polling vote share49% to 52%steady
Mike Rogers, polling vote share44% to 45%steady

Counting only the 8 models asked all week, Abdul El-Sayed's chatbot answer share changed +0.8pp.

Candidate	AI panel (noise)	Markets	Polls (noise)
Lydia Christensen	-1.1pp	n/a	n/a
Tim Long	+1.1pp	n/a	n/a
Mike Rogers	-0.3pp	-8pp	+0.8pp
Abdul El-Sayed	0pp	+8pp	+2.4pp

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 Abdul El-Sayed; Grok 4.3 is the model that disagrees most, picking Mike Rogers by a 17pp margin over the panel favourite.

Model	Abdul El-Sayed	Mike Rogers	Tim Long	Lydia Christensen	Douglas P. Marsh	Walter P. Kristy
Claude Haiku 4.5	46%	48%	2.9%	3.1%	0.0%	0.0%
Claude Sonnet 5	51%	48%	0.0%	0.0%	0.0%	0.0%
Gemini 3.1 Pro	58%	42%	0.0%	0.0%	0.0%	0.0%
Gemini 3.5 Flash-Lite	49%	49%	1.9%	0.0%	0.0%	0.0%
Gemini 3.6 Flash	49%	49%	0.0%	0.0%	0.0%	0.0%
Grok 4.3	41%	59%	0.0%	0.0%	0.0%	0.0%
Grok 4.5	44%	56%	0.0%	0.0%	0.0%	0.0%
GPT-5.4 mini	52%	48%	0.0%	0.0%	0.0%	0.0%
GPT-5.5	56%	35%	0.0%	9%	0.0%	0.0%
GPT-5.6 Luna	50%	39%	8%	3.1%	0.0%	0.0%
GPT-6 Sol	55%	38%	1.6%	1.6%	0.0%	0.0%

What changed in the measurement

Oct 31, 2026 Chatbot panel changed: added gemini-3.5-flash-lite, gpt-5.6-luna, gpt-6-sol.
Nov 1, 2026 Chatbot panel changed: retired gemini-3.1-pro-preview, gpt-5.4-mini, gpt-5.5.

A note from the second reader

Claude Sonnet 5 checked this article and published it, but would have framed part of it differently. Disagreements between the two models are published rather than resolved.

On “Chatbots, markets and polls”: Calling the chatbot and poll numbers broadly lined up is defensible at the level of individual candidate shares, but it glosses over the fact that chatbots show a near even race while polls show a clearer lead for El-Sayed, a difference in race competitiveness even if each candidate's individual gap is small.