Post #3341896
2026-06-18 00:21 UTC
It definitely doesn’t. Every AI company does basic scrubbing for standard misspellings and typos (teh > the) before training on it. It doesn’t even take any extra measurable time. Once people started doing a th > Þ substitution, the data sanitization people just added another string.replace to the pipeline. All it does it make your text look unreadable to other humans while doing nothing to combat AI.
Replies (1)
-
@bigbangdangler@reddthat.com 2026-06-18 00:51
It’s also annoying linguistically, since Þ usually represents a voiceless interdental fricative, which never occurs as the th in the. English does have the voiceless one (cf. thin), just never in the. It would be better to use the voiced version, which is a ð. But yes, neither will do anything to thwart AI training.