ChatGBT

[BLAZE]

I mentioned on a thread the other day that the spambots that visit this site are becoming eerily more convinsible. Along a similar line, has anyone tried the recently released ChatGBT? If you dont know what it is, this is worth watching;

https://youtu.be/kh5dN72GTQ8


I cant help but wonder where humanity is heading here. Are we in danger of making ourselves obsolete?

How can the education establishment police plagarism, when a student is able to generate an entirely convinsible essay is 3 seconds flat? We're not quite there yet but it isn't far away

And what about artistic expression when AI is able to generate images, music and literature that is indistinguishable from that of man? We're virtually there (Google AI generated artwork)

How long until the first artificially intelligent football manager, judge or government?

Strap yourselves in lads and stock up on the tin foil

 

[Neath_Jack]

Can someone show this to Russ please.

 

[Jack2jack]

Well if it all goes belly up, we'll only have our selves to blame. Too smart for our own good sometimes.

 

[BLAZE]

https://www.dailymail.co.uk/sciencetech/article-11750405/ChatGPT-insulting-lying-gaslighting-users-unhinged-messages.html

 

[Skippyjack]

Lord Bony would have been all over this stuff, mad tw*t that he was

 

[MrSwerve]

https://www.planetswans.co.uk/forum/viewtopic.php?f=2&t=11662

 

[MrSwerve]

I cant help but wonder where humanity is heading here. Are we in danger of making ourselves obsolete?

How can the education establishment police plagarism, when a student is able to generate an entirely convinsible essay is 3 seconds flat? We're not quite there yet but it isn't far away

We certainly are there, and are seeing it already in our University. It's incredibly hard to police. Our plagiarism detection (Turnitin) doesn't detect any of this stuff.

We've noticed however that ChatGPT does make up some references that do not exist - so it's worth checking reference lists (I now do it in more detail than I used to!).

There are systems being developed already that are designed to pick up on potential AI-generated documents, but they are by no means 100% accurate. These will be integrated into University systems 'soon'.

Just as worrying for us is the fact that it can generate workable programming code with no real way to tell if it's AI generated, especially if the student then changes variable names etc.

There are plenty of staff with us that want us show students to harness AI rather than avoid it, I'm on the fence about this to be honest, or we'll have a generation of graduates that don't know the actual intricacies of how systems work. It's a bit scary really.

On top of this, most Universities (probably all Universities at this point) don't have policies on this. Technically asking an AI to generate something from an input from a student isn't classed as plagiarism. Sure, it's not something that they've typed out but its neither something that has been copied directly from elsewhere. The AI can also regenerate sentences until you are happy with it. It really is a minefield.

 

[BLAZE]

I just noticed something else that's blown my mind. This spambot - https://www.planetswans.co.uk/forum/memberlist.php?mode=viewprofile&u=5086

Posted in this thread - https://www.planetswans.co.uk/forum/viewtopic.php?f=3&t=1960&p=158581#p158581


Not only was it intelligent enough to understand a game was being played, it also worked out the rules of said game and posted a valid contribution to it. Other users interacted with it none the wise that it was an artificial intelligence

 

[BLAZE]

Some of the benefits

https://www.businessinsider.com/ai-can-detect-breast-cancer-better-than-doctors-study-says-2023-3