AI | Frank Titze

Where I use AI?

For text generation and translations to speed up those things.
For picture analysis for content including keywords creation with focus on getting at least good drafts.

Where I do NOT use AI?

Why do I use AI?

Simple answer: In mid 2013, I considered to create picture descriptions. But pure the creation of 13,000 picture descriptions (status mid 2025) in one language would have cost me likely one (1) year full time work. Not counting the translation to my second used language.

Where my/the AI (usage) is limited?

The AI generated results are often a good draft but neither perfect or always correct. I guess an aproximate 90% correctness.

How do I overcome with the AI limitations?

For non direct to picture related texts, I take the AI "draft" and review and modify it. My focus here is on avoiding incorrect content identification (e.g. cathedral vs. minster) as well as the length of the descriptions or generated titles. Otherwise, the descriptions and ratings remain unchanged.
For descriptions and keywords of pictures, I do a quick review of the AI generated texts and edit what is too misleading or wrong. Doing so expect to improve the aproximate 90% correctness to 97+%.
Whether the AI generated texts are reviewed or not is indicator with an AI icon in green or orange:

	Text is AI generated - Reviewed / Corrected - Approximate 97+% guessed correctness.
	Text is AI generated - NOT reviewed and/or corrected - Approximate 90% guessed correctness.
	No AI text generated