Interesting to hear these experiences. I'm sure there will be a period now when people discover further and further use cases. It seems to be a revolutionizing tool.
I replaced google with chatGPT and it's really good to help with issues you have in your work. As an example our IT department stopped us from installing eclipse plugins from the marketplace and I was just lost getting no help from googling. So I just asked ChatGpt how to install gradle without using the marketplace and got a perfect answer that was of such good quality that I pasted it into our dev guides. If you want to know how to do something you get a good start by asking ChatGpt but it's not that good at writing code yet. Asking it for code review of small code portions is also useful. It's also fantastic at explaining how things work and a very good tool for learning stuff. Learning how to use ChatGpt to improve your code quality and productivity is how we should approach it. And the more experienced you are, the more effective use you will have of it.
On Sat, Mar 18, 2023 at 10:03:15AM +0100, Mikael Djurfeldt wrote:
[...]
> > I know very little apart from knowing what deep learning is and having
> > skimmed the "Attention is all you need"-paper.
That's already more than I do, I think. Thanks for the pointer :)
> > I only meant that you are
> > not training the model during and between sessions.
No, I don't think we are yet at "real time model training". But who knows.
> > It is certainly
> > possible that OpenAI filters out things from the dialogs to use as part of
> > training for the next version. They warn you that they may take data from
> > the dialogs. If and how they do that I don't know.
I'm pretty sure that this is the plan (besides, of course, of trying to
assess, understand public reaction to the whole idea, and perhaps to
steer public perception, as far as possible).
> Or, as GPT-4 would phrase it: I apologize for the confusion in my previous
> answer. You may be right that I'm training the next version of the model. :)
:-D
Cheers
--
t