When I ask anyone/anything a fact-based question, I expect an answer that isn't embellished or followed by a question that solicits information from me.
Here's a great example: I asked "What's the tallest tree in America?"
- Bing: "Hyperperion, according to two sources." It then showed the appropriate paragraphs from Wikipedia and from the NY Times.
- Google: Listed the same Wikipedia article. However, before that, it showed a Gemini response that was wrong: "The tallest tree in the United States is the Doerner Fir, which is 327 feet tall and located in Redwood National Park in Oregon. It's the second tallest tree in the country and the world's tallest Douglas fir. " A second Gemini paragraph listed the correct tree as the tallest in the world.
- New Copilot: "The tallest tree in America is Hyperion, a coast redwood (Sequoia sempervirens) located in Redwood National Park, California. Hyperion stands at an impressive height of 379 feet (115.5 meters)1. Its exact location is kept secret to protect it from potential damage by visitors1.
Have you ever visited any of the redwood forests? "
- Gemini (not from Google search): "The tallest tree in the world, and therefore also the tallest tree in America, is Hyperion. It's a coastal redwood (Sequoia sempervirens) located in Redwood National Park, California. As of 2019, it stands at a towering 380.8 feet tall! Hyperion is not only the tallest tree in the world, but it's also one of the oldest, estimated to be between 600 and 800 years old."
- ChatGPT: "The tallest tree in America is a coast redwood (Sequoiadendron giganteum) named Hyperion. It was discovered in 2006 in a secret location in California's Redwood National Park and stands at about 379.7 feet (115.7 meters) tall. This makes it not only the tallest tree in the U.S. but also the tallest known living tree in the world!"
The responses are all correct (except the Gemini summary stuffed info the Google search response), but Copilot wants to engage me in chat. If I answer yes, it wants to know which redwoods I've seen. It's like Copilot has scraped all the content it could get from the web and has now decided to start scraping its users. Ick.