Or to offer use of models that explicitly state no training / fine tuning (usually they say ‘improving’) with sent prompts / responses. Google Gemini has t&cs along those lines for API use, I bet others do too.

If in doubt, unless you see written policy otherwise, assume your data will be used by OpenAI to train their models.

Looks good but just to advise some caution - because it is using commercial inference providers like OpenAI there’s no transparency there about the downstream t&cs about using your data and the responses for training or fine tuning those commercial models. And those vary from provider to provider.

Yes! There are loads of open models with very promising results and huge potential for collections data work. I can list a bunch here if it’s useful or we could talk on Teams some time… I am about to start experimenting with them on our new NVIDIA H-100

Saw it last night and loved it - thank you for the steer!

Thanks - I was debating whether to give that a try, I’ll do it

This is so useful James, thanks very much! I am going to copy your methods…..

James I’ve not done much in the way of lists and feeds yet, what would you recommend as an approach to it?

Oh wow! I need to look at more daschund photos obviously

Beautiful spaniel! Sprocker? Here’s our Sprocker Truffle

That record changed my brain. RIP.

And here's a video! I know, moving image innit www.youtube.com/watch?v=TOUe...