thisorthis.ai

The whole interaction is a single choice. You tap a logo, optionally tag the job you use AI for and the models you run, clear a reCAPTCHA check, and you're done. No account, no password, no email list. The weekly cap keeps one person from stuffing the ballot while leaving the barrier low enough that casual visitors actually vote.

The main AI leaderboard ranks by share of this week's votes. A tab switch moves you from platforms to individual models, which is the more interesting view once you already know you want a Claude or a Gemini. Both tabs carry columns for change against last week and last quarter, so a single strong week doesn't read like a trend.

A world map colors each country by its leading AI, and a country needs at least two votes before it gets a color. Below the map, a list ranks countries by vote volume and names the front-runner in each. It's the quickest way to see where ChatGPT holds ground and where Claude has pulled ahead.

Repeat and automated votes get filtered. The site combines a browser cookie with network signals and human checks, and each vote passes a reCAPTCHA gate. Votes that look scripted or duplicated are dropped rather than averaged in. That matters for a poll this small, where twenty fake votes would rewrite the table.

Voters can say what they use AI for, and the site aggregates those answers into a top pick per job: coding, writing, research, video, everyday chat, marketing. That turns a popularity contest into something closer to a use-case read, with vote counts shown next to each winner so you can judge how thin the evidence is.

Five explainers sit beside the poll: which AI to use for a given task, how AI works, how to write prompts, an A to Z glossary, and a history of AI from 1943 onward. They're bylined and reviewed with dates, and they carry the same argument as the leaderboard: no single AI wins every job.