The Technium

Lie Detection


There is no absolutely infallible way to detect a lie in the present. We can only reliably detect lies retrospectively, in the frame of history. So the only way to not be deceived in the present is to rely on sources that have been proven true in the past. The more truthful a source has been in the past, the more reliable (on average) they are in the present. Therefore the only method for detecting fakes in the present is to depend a source we trust over time. This method is not infallible (see first point) but it is the best we have.

This type of trust applies to text and images. There is no way to detect a lie by simply looking at words in a sentence; they may or may not be what was actually said, or they could be altered. You can’t tell by looking. There is no way to detect a lie in a photograph or video simply by looking at it either; it may or may not have been photoshopped or altered, or generated by AI. We must ultimate trust the source that distributes them. That kind of hard-earned trust can be easily damaged. Sometimes individual people with long histories can be trusted. More often we trust institutions with long histories and many individuals, such as some government agencies, science associations, journalistic companies, non-profits, and sometimes corporations. We tend to trust them — until they are caught lying. When this happens it’s a blow to our sense of security, because there is no absolutely infallible way to detect a lie in the present on our own.




Comments