Have you played with OpenAI's ChatGPT?

[Lost User]

Quote:

Moscow experiences a wide range of average high and low temperatures throughout the year. Based on information from the National Oceanic and Atmospheric Administration (NOAA), the following basic summary of Moscow's average temperatures is provided:

[Spinbot - Article Spinning, Text Rewriting, Content Creation Tool.](https://spinbot.com/)

"Before entering on an understanding, I have meditated for a long time, and have foreseen what might happen. It is not genius which reveals to me suddenly, secretly, what I have to say or to do in a circumstance unexpected by other people; it is reflection, it is meditation." - Napoleon I

[jmaida]

ChatGPT draws on a lot sources. No surprise.

"A little time, a little trouble, your better day" Badfinger

[RainHat]

I like the was this totally stops the machine. 1. Stop all interrupts. 2. Enter into an infinite loop. Worst code possible.

[DerekT P]

Yes, a little. In some cases the answers seem very good, in others they can be incomplete, contradictory, or just wrong. When errors are pointed out, ChatGPT sometimes improves the answer (and sometimes apologises), but it seems to me it generally bases its responses on the first thing it looks up, rather than comparing a number of answers and finding a consensus. When I asked "whats the difference between walschaerts and stevenson valve gear?" (types of valve gear used on steam locomotives) it gave a clear answer, describing not just the differences but the advantages/disadvantages and why you might adopt one over the other. It felt like talking to a knowledgeable engineer with good communication skills. But when I asked "how do vacuum brakes work on a railway train" the initial answer was poorly described and included some confused and misleading descriptions. When I asked it to clarify an aspect it had pretty much fudged, it gave me an equally poorly worded response that said pretty much the opposite of what it had just said. I challenged it on its contradiction and it apologised, admitted it was wrong and gave a third poorly-worded spiel that didn't clarify anything and still used wrong concepts and omitted key aspects. As with (for example) newspapers, we take information and tend to assume it's true. But when it relates to a field we're knowledgeable about we quickly spot the flaws and inconsistencies and have to assume that the source is just as uninformed about other subject matters, and that we're being fed complete nonsense a LOT of the time.

Telegraph marker posts ... nothing to do with IT Phasmid email discussion group ... also nothing to do with IT Beekeeping and honey site ... still nothing to do with IT

[Nelek]

DerekT-P wrote:

As with (for example) newspapers, we take information and tend to assume it's true. But when it relates to a field we're knowledgeable about we quickly spot the flaws and inconsistencies and have to assume that the source is just as uninformed about other subject matters, and that we're being fed complete nonsense a LOT of the time.

Sadly many many people doesn't see it. And it gets even worse, when the one who explains bullshit is someone famous in social media.

M.D.V. ;) If something has a solution... Why do we have to worry about?. If it has no solution... For what reason do we have to worry about? Help me to understand what I'm saying, and I'll explain it better to you Rating helpful answers is nice, but saying thanks can be even nicer.

[Kornfeld Eliyahu Peter]

The interesting thing about this chat AI, is how it evolves... At the end of our discussion I gave it a nice solution (actually the BELL form the C128, a well proved code). The AI acknowledged that it is a good solution (based on what?), but two days after still offering silly code when I ask the same question...

"Everybody is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid." ― Albert Einstein

[Nelviticus]

I believe the point of ChatGPT is not to give correct answers but to sound like the answers have been written by a real person. The fact that it's sometimes also correct is just luck.

Regards Nelviticus

[MonkeyVaughan]

I'm a software engineer, mostly contract stuff so I like to keep learning the latest technologies and stuff. I'm completely blown away by ChatGPT's possibilities. I don't feel like it will put me out of a job but it will make my work life easier, for example when I was testing what it was capable of I got it to write me a C# .Net MAUI, cross platform mobile speech to text interface and implementation, RESTful WebAPI calls and text to speech interface for responses, including dependency injection, unit tests and saving conversation snippet history to a SQLite database. Pretty blooming impressive! Some of the code it produced isn't perfect but it is a shed load better than google and even StackOverflow. A large part of our job as software engineers is knowing how to ask the right questions and being able to sort the wheat from the chaff but watching it go creating about 3 or 4 days work in about 20 minutes that saved me a lot of typing really blew my mind. To show my friends and family that I'm not completely obsessed by the whole "The future has arrived!!" vibe I was giving off I've also got it to write very funny screenplays for scenes of their favourite TV shows such as Friends, Charlie's Angels, The Big Bang Theory incorporating them in the storyline and the screenplays it produced were astonishing, kept really well within the style of each show and their characters. You can give it a theme/style for a chef's menu and it can produce one all formatted and provide recipes. It is only limited by your imagination for asking the right questions and giving it the inputs to come up with answers. It's a bit scary but amazing.

[Lost User]

MonkeyVaughan wrote:

It's a bit scary but amazing.

The only thing I find scary about it is that it can mimic humans and go undetected. I think ChatGPT looks useful for determining what a human is asking about. In other words, I see an immediate use for search engines and Q/A websites.

[Damien Hansen]

I think that AI-powered language models like ChatGPT can be used to assist with tasks such as language translation and language learning, which can make it easier for people to communicate and access information from around the world. :thumbsup: