Chat GPt - a way to go for reliable information

[mikem]

Hade is actually the angle of inclination of a fault / rift / lode, rather than the feature itself.

Sorry, yes it's a feature of the rift, but that is already implied in the word.

Graph of uses of hading from 1800 to present:

https://www.google.com/url?sa=t&source=web&rct=j&url=http://books.google.com/ngrams/graph%3Fyear_start%3D1800%26year_end%3D2019%26corpus%3D26%26smoothing%3D7%26case_insensitive%3Don%26content%3Dhading&ved=2ahUKEwiQ4Pfbq9L-AhVNg1wKHVPrDV4QhCx6BAgOECo&usg=AOvVaw0h76JNIciUL1SjUhd4_4kO

 

[bills]

Course it can do calculus as most of calculus is a set of basic algorithms!!

 

[mikem]

Yes, but does it, or does it look for answers first?

 

[bills]

A good point, but as an engineering geologist rather than a mathematician I have relied on Schum's handbook of Mathematical Identities and the like for 40 yrs, without feeling the need to prove each one. I think it is about a matter of trust, checking and use where appropriate, but if I can use something that is going to give me the right solution efficiently then I do not see why it should not be used as a tool like any other piece of computer software.

 

[alanwsg]

I was trying to solve a puzzle and asked it ...
Give me a 9 letter word that means rotting or decaying

The reply ...
One possible word that means rotting or decaying and has nine letters is "putrefying".

 

[2xw]

A lot of these comments are the same as complaining that your microwave hasn't cleaned the bathroom

 

[aricooperdavis]

A lot of these comments are the same as complaining that your microwave hasn't cleaned the bathroom

Absolutely. ChatGPT isn't trying to be a generalized artificial intelligence, but a language model. The fact that it feels like a generalized intelligence says a lot about what we consider intelligence to be.

 

[ChrisB]

On another forum I moderate (not caving related) we require new users to have their first post approved, to eliminate spammers. We are now getting very plausible replies to posts, written in the style of ChatGPT. The other give-away is that the threads replied to are 5 to 10 years old (they hope nobody notices) and although it's a UK forum, the IP is somewhere in Asia and the email address is typically in Oz. But if you didn't know about ChatGPT, you'd think they were from a real person. It's only a matter of time before they improve to the point where you can't tell.

PS - yes, we do have a captcha, but I think AI is solving that too

 

[2xw]

On another forum I moderate (not caving related) we require new users to have their first post approved, to eliminate spammers. We are now getting very plausible replies to posts, written in the style of ChatGPT. The other give-away is that the threads replied to are 5 to 10 years old (they hope nobody notices) and although it's a UK forum, the IP is somewhere in Asia and the email address is typically in Oz. But if you didn't know about ChatGPT, you'd think they were from a real person. It's only a matter of time before they improve to the point where you can't tell.

PS - yes, we do have a captcha, but I think AI is solving that too

A month or so ago researcher's tasked one of these models to solve a captcha and it went on taskrabbit and hired a human to do it by telling them it was a person who was blind and couldn't solve captchas.


There are other AIs that can detect chatgpt output but obviously I can see it adding to your burden as a moderator!

 

[pwhole]

There are other AIs that can detect chatgpt output but obviously I can see it adding to your burden as a moderator!

Just look for all the post's with the inappropriate apostrophe's, and they'll be the human ones. 'No HGV's'

 

[mikem]

But it also copie's our response's...

 

[ZombieCake]

AI seems to be a very bad thing if not handled correctly. It looks like the Americans have just invented Skynet for real. Wonder if the drone operator was called Sarah Connor?

US air force denies running simulation in which AI drone ‘killed’ operator

Denial follows colonel saying drone used ‘highly unexpected strategies to achieve its goal’ in virtual test

uk.yahoo.com

 

[RobinGriffiths]

Looks like Asimov's Three Laws of Robotics was the first thing to go out of the window.

 

[mikem]

The same Colonel also says it didn't happen, see simulated test link:

US air force denies running simulation in which AI drone ‘killed’ operator

Denial follows colonel saying drone used ‘highly unexpected strategies to achieve its goal’ in virtual test

www.theguardian.com

 

[aricooperdavis]

I appreciate I'm being a bit anal here, but I think it's worth reiterating as this is very much in the public consciousness and a source of much discussion (and scaremongering) right now:

ChatGPT is not an artificial intelligence, it's a language model. It's not piloting drones, winning games of chess, or dreaming of electric sheep. It's just very very good at guessing what word(s) are likely to follow others.

 

[mikem]

It may not be, but it is from a company potentially working towards that end (from their website): "Today’s research release of ChatGPT is the latest step in OpenAI’s iterative deployment of increasingly safe and useful AI systems."

 

[aricooperdavis]

I don't think OpenAI will produce an AGI. I'm not convinced it's possible, or particularly useful, and they're certainly not trying to develop anything like one at the moment. However I think their language models have the potential to be seriously disruptive to the labour market, and we're not at all prepared to deal with that very real and current risk.

 

[Graigwen]

ChatGPT is not an artificial intelligence, it's a language model. It's not ... dreaming of electric sheep.

Good allusion!
.

 

[RobinGriffiths]

Ooops. Replied to wrong thread....

 

[Graigwen]

What answer is provided if you ask ChapGPT "How many entrances does Ogof Draenen have?".