What did I do this weekend?
I administered a pop quiz to AI 🤓 🤖
Warp recently launched a feature called AI Command Search. It takes natural language descriptions and converts them to terminal commands for you in the matter of seconds.
But the biggest hesitation I had was "How reliable is this command & can I actually trust what it outputs?"
In my experiment, I used a cheat sheet with 24 of the most commonly used git commands. I took the natural language descriptions of these commands, put them into Warp's AI Command Search, and scored the AI based on how well the output matched what was on the cheat sheet. The highest possible score was 24 points, for the 24 commands on this cheat sheet.
Warp's AI Command scored 22.1/24, for a score of 92%!
If you're curious about how I scored everything, you can read the blog below 👇
https://www.warp.dev/blog/replace-git-cheat-sheet-ai-command-search
Top comments (0)