Why AI really is not ready for prime time.

[Medlin Software, Dennis]

I saw and objected to such responses on another private board. The person eventually responded stating they thought they were helping (they would copy the OP, then post AI results, which was easy to reproduce). I was blunt, and said the board was for providing "between the lines" sharing, as all of the participants should be professional enough to use a search engine, so asking in public was almost never for a plain answer any search engine could return. Of course, there are many newer members who ask things, but most of the vets ignore things which can be found via even the most basic search.

Search engines are the precursor to AI, since they scrape and snip things to present. The difference is, at least for me, being in a search engine results in earnings, where AI results to not. I allow one or two search engine AI bots "in" at present, since they give me something back (more search engine click throughs). One of the first bots I stopped was one who was selling plagiarism score results. I got nothing from them hammering my sites, yet they were making money. I also blocked a competitor's IP ranges, as they were hammering my site as well. There is a great long time group for web site creators to discuss and share these types of things (I don't work at it that hard).

The last year or two, more and more random IP's are scraping me looking for doors in. Most look for common word press pages. Easy to block though. This is another reason I do not use tools like word press. Better to be a little fish in a big pond, rather than invite hacking. We use some shared type web coding, but we do not depend on outside tools other than certain well known and trusted css files. At some point, I will likely have to create a door in, meaning washing all accesses through a filter type of deal. Many use Cloudflare for this, but I prefer something less well known.

[jklcpa]

52 minutes ago, Medlin Software, Dennis said:

they would copy the OP, then post AI results, which was easy to reproduce

This is exactly what drew my attention where this newer member would regurgitate the OP question and facts, and then incorporate an answer. It seemed weird and robotic because the tone here and on many forums these days is more casual and conversational. 

 

[Medlin Software, Dennis]

I get accused of being an AI bot. The issue is not my reply, but the number of same human tricks, needing the same response. I do have prewritten replies I can select, honed to be as concise as possible, but they were human developed/written.

Last week, it was a person who was having trouble following plainly and accurate steps (three of them). Their replies did not need any more from me, other than to remind them to accurately follow steps in the earlier reply. After a few messages, I added something like "The first reply had the accurate solution, step by step. There are no hidden or additional steps I held back." This resulted in the oft seen "I want a real person you must be replying with AI." response.

I sometimes use the old "paper towns" trick - not correcting all spelling errors, or even insert one, to not only see if anyone reads the item, but to now prove it was not AI generated.