Correct text & formatting function (AI)

[Alpha1]

It regularly happens that members post a wall of text with poor grammar. i.e. no line breaks, bad spelling and formatting mistakes. This is hard to read and sometimes completely unreadable.

It would be nice and helpful to have a function to check the text and suggest new spelling and formatting. For example by a editor button / link the member can click.
AI could suggest text improvements, which the member may want to apply.

And for moderators this can also be very useful to be able to quickly turn an unreadable message into a readable one.

 

[Lee]

While not against this, all these AI service calls are going to get expensive, no?

 

[Alpha1]

In general LLM calls can get expensive, but not for this feature.
The call would only be made if a user clicks on it. It would only send the text of the post, so its quite limited.

 

[Lee]

Yeah, I understand.

I'm just curious, if all these AI features ever come to light, the monthly cost of running a forum is going to increase quite a lot isnt it?

 

[Suzanne O]

No thanks. Don't want my forum run by artificial intelligence robots. They get misused too much and can wreck your database.

 

[Alpha1]

Features will always be optional and up to the administrator to turn on or not.

There are other suggestions I can think of that would be expensive and therefore I'm not posting those. I do expect that llm calls will become much cheaper over time. AI will also open up new opportunities to monetize forums, so it works both ways.

I'm already using AI text formatted functionality for free by combining several free services and manually converting the texts into it.

 

[Lee]

Yeah I agree with all of the above.

Yeah, i have written myself an addon to scan all user submitted content for illegal harms.

Send me a message for your ideas and I’ll see if it’s something I’m interested in playing with.

I am still learning though so progress may be slow.

 

[Alpha1]

Yeah, i have written myself an addon to scan all user submitted content for illegal harms.

Scanning all content will indeed be costly at this point in time. I think that when scanning mass content, its probably best to then get all the relevant functionality to save cost. i.e. correct title, prefix, tags, correct thread type, illegal harms, rule breaches, formatting, related content, ads, etc, etc. Else it would become costly to scan content for each purpose, again and again.

 

[Lee]

Again, I absolutely agree with this statement.

I only need a harm scanner in the short term but there is no reason this can’t be expanded