Forum: Author Hangout

Why is it ok to use text to speech tools that are also AI but not an LLM to edit?

Possibly, because text to speech just verbalises what is written, whilst an LLM editing actually changes what has been written. Sometimes drastically. Those changes are not changes by the author, they are (technically) changes by another individual, which makes the edit not of the original authors voice.

I think you don't know how to use an LLM then.

My ability to use, or not use a LLM does not matter, as that is not the question you asked.

You asked;

Why is it ok to use text to speech tools that are also AI but not an LLM to edit?

I replied with the most obvious answer. The important words here are "text to speech" and "Edit". Those are two entirely separate processes/mechanisms. If you fail to grasp that, then I think you are failing to effectively vocalise the question you actually want the answer to, and until you do, you will not receive the answer you are seeking.

You specifically mentioned that an LLM changes content versus a speech to text AI just verbalizing content. These are your exact words

No, no they are not my exact words. Not even close.

Sounds like verbal gymnastics to me because you are ok with one but have an incorrect understanding of another process.

You are still failing to understand the issue. In fact, I would even go as far as to say that you are overcomplicating it. I don't know how to answer the question you are asking in a from you can understand. I shall, yet again, quote your original question so as to not get sidetracked.

Why is it ok to use text to speech tools that are also AI but not an LLM to edit?

PROCESS ONE: Uses an AI/LLM to take a submitted piece of work and verbalises it exactly as it has been originally written. No alterations, no edits, no paraphrasing, no summary, no opinion based on inherent bias,no coffee black no sugar. This is okay, as the original piece has not been changed in any form other than to be transferred from type media to audible media.NO EDITS HAVE BEEN DONE, so it doesn't 'matter' if AI or a LLM was used.

PROCESS TWO: Uses an AI/LLM to take a submitted piece of work and adjust it to a narrative bias that has been intentionally or unintentionally, installed during the LLM's/AI's programming. This may include adding content that never existed, or removing content that existed. Both can have dramatic effect on what was originally said and what was originally meant. (For example, what the BBC did to Trump's speech). It's not OK to use an AI/LLM to edit because you don't know the bias of the programming used to create the AI/LLM (another perfect example was early AI/LLM's insisting that all Vikings were black, because that was the bias of the programmer) AI's/LLM's have bias and when you use them, you are installing that bias in your work. Also people who use AI/LLM for content creation, is it really their work? It's their idea with anothers input. Very much like any press release ever, umm, released.

So, to summarise. AI/LLM used to change text to audible, changes the content from visual to audible, that's it. AI/LLM used to edit, takes the content and changes the shit out of it, and not always in a readable way. This is why no one cares if AI/LLM is used to turn written words into sound and why people care about AI/LLM being used to write/edit content.