How often the AI assistants agree on the same recommendation
A running measure of whether ChatGPT, Claude, Gemini and Perplexity converge on the same top pick — or send travellers to different places.
As of October 14, 2026, the AI assistants agreed on the #1 recommendation 11% of the time — across 7,437 questions where at least 3 of the 4 assistants gave a clear top pick, spanning 124 tracked destinations.
A "comparison" is one tracked question where at least 3 of the 4 assistants returned a clear top recommendation on their most recent run; it "agrees" when every one of them named the same place. The quorum (rather than "all 4") keeps a provider outage from zeroing the metric — it shrinks the sample instead. How we track this.
The tracked trend
Last 7 checks · holding roughly flat, no sustained convergence. This series is the stricter all-4 measure frozen at each check close, so it reads lower than the headline quorum figure — it's the line to watch for direction, not the absolute number.
Which assistants answered
How many of the 7437 comparable questions each assistant actually gave a top pick on.
Where the assistants disagree most
Tracked destinations with the lowest cross-model agreement (at least 4 comparable questions each).
- Sherbrooke 0% agree
The assistants landed on the same top pick in 0 of 60 comparable questions.
- Mississippi 0% agree
The assistants landed on the same top pick in 0 of 60 comparable questions.
- Medicine Hat 0% agree
The assistants landed on the same top pick in 0 of 60 comparable questions.
- Brandon 0% agree
The assistants landed on the same top pick in 0 of 60 comparable questions.
- Iowa 2% agree
The assistants landed on the same top pick in 1 of 60 comparable questions.
- New York 2% agree
The assistants landed on the same top pick in 1 of 60 comparable questions.
- North Carolina 2% agree
The assistants landed on the same top pick in 1 of 60 comparable questions.
- Canada 2% agree
The assistants landed on the same top pick in 1 of 59 comparable questions.
Why this matters
If you ask several assistants the same travel question and get a different answer from each, no single one of them is really "the" AI recommendation for that destination — and a place that tops one model's list can be invisible on another. Tracking convergence over time is how you tell a stable AI consensus from noise, and it's the number most likely to move as the models retrain and change how they search.