Someone is feeling a need for publicity…
Peter wrote:
I don’t think the world will be worth saving, because the IQ will decline to zero all by itself, as a result of massive false eyelashes blocking out sunlight
I thought Rishi Sunak was an intelligent person enough to employ advisers with some knowledge in the area if not an experts. So either I was wrong or this is completely fabricated by journalists.
Or fabricated halfway up the information path. Right now everybody is working hard to look “right on” and organisations have whole departments working out how to do this.
But we know this is BS and nothing will actually happen. So it should be harmless. And politicians are experts at kicking stuff into the tall grass, as the saying goes.
We are getting more ChatGPT posts on EuroGA. I removed one signup yesterday. But so many signups are bogus anyway – this is a recent example
5 real, 7 malicious. So dealing with ChatGPT posters is all in the same job 
Such a change from the early days, Oct 2012. Back then you could sign up and post right away, and under 1% would be malicious. Some were truly nasty though. But we ran with that for a few years. ChatGPT brings in a new dimension because if the mod(s) is not on the ball, the twat could generate many posts before anybody spots it. Especially as ChatGPT is carefully done to avoid expressing an opinion on anything whatsoever so nobody can possibly object to anything it says.
Ha! I have an idea. Delete all non-opinionated posts? That would be an interesting captcha on the signup page. Ask about some current non-PC issue. A multiple choice question on the IQ of Holly Willoughby, with options of 0 10 20 30 40 50 50+ and anybody choosing 50+ is not allowed in
You could ask “How do you impersonate a Frenchman?” because ChatGPT will refuse to tell you how to do that! :-)
All you pilots – got to be extra careful these days
Peter wrote:
All you pilots – got to be extra careful these days
Daily trash goes beyond imagination
I am sure that article is what the writer actually did. I wouldn’t go on a date with her even if she was going to give me a TBM.
The amazing thing is that somebody would actually do it
Again this is what ChatGPT does. If you are looking for a date, it will trawl that 2021 internet archive for matching phrases. The result will fool an awful lot of people.
AI? Not at all.
But somebody put in a lot of man-hours to collect up a load of fairly specialised www material. I know somebody who used ChatGPT to get off a load of parking ticket fines; it had some chatbot which asked questions on the circumstances etc and generated a letter to send to the court.
And this has been expressed as a significant risk with using ChatGPT or similar tools: the text it generates might be recognisable and in narrow contexts might reveal to somebody what you were looking for.
It’s somewhat similar to the old dropbox security vulnerability: it hashes each candidate uploaded file, and if you are uploading a file with an identical hash, it doesn’t actually upload it; it just pretends to, and the upload is very fast. So if a company wants to check if one of its employees, or anyone worldwide, has uploaded some internal document to dropbox, it just uploads it and sees how fast it did it.
A year or two ago a developer of early chatbots was deceived by one of his own creations on a dating site. Frustratingly I now can’t find the article to prove it.
I remember someone at university memorising lists of trees in German. The funny thing is she knew that die Ulme = elm, but didn’t have a clue what one actually looked like.
ChatGPT isn’t very good at maths (the one thing that computers should be good at). Spot where ChatGPT got it wrong (a real schoolboy error).
Compare with the same question asked of Wolfram Alpha:
