The accompanying web browser Dillo has proposed a method to filter out changes made through AI. The Dillo project allows patches created only by humans, but reviewing submitted changes is time-consuming and it's not always clear immediately whether a patch was created by a person or not. To simplify filtering out AI-generated patches, first-time contributors submitting changes to the project are asked to provide a recording of the development session as proof of their work. When using Vim, the session can be recorded, for example, using the asciinema utility.
The idea has not been converted into mandatory requirements and is still under discussion and experimentation. At first glance, a developer trying to deceive the maintainer by presenting an AI-created patch as their own code could use AI to generate a session recording in asciinema as well. Experiments have shown that the artificiality of such recordings is immediately noticeable, as they do not reflect the typical human actions, errors, and thoughts during the work process. Standard AI models are trained on already prepared code, but their training does not fully cover the process of coding by humans.
Source: opennet.ru
