Chat GPT and Christmas Turkey

 Chat GPT and Turkey Stuffing

And not sure if this should be in my work blog or my recipe Blog. 

I wanted an apricot stuffing recipe for a Christmas turkey and found that most recipes Google came up with had sausage meat or mince of some kind, so I turned to Old Mate Chat GPT.

So, first my prompt:

One webinar I attended suggested being polite and courteous to the tool as we are training it to interact with us and if we want politeness and friendliness we should demonstrate that (and yes I know its just a bit of code).  I also learnt that if you say 'I need you to try hard' you do actually get better results.  

And then this recipe popped up within seconds:

Followed by instructions:

Now this looks like it could work, but look at these pleasantries - I made Old Mate "incredibly happy":

Now, I'm a cook and while this looks like it might actually work I had a suggestion and a query:

Which resulted in a  

And a revised recipe: 

I'm going to make this recipe because I think it's probably going to work but it got me thinking about a few things:

• recipe development - one of my favourite cooks at the moment is Justine Doiron who is not a trained chef and makes wonderous recipes using combinations discovered through experimentation. If AI creates recipes that work based on what its learned from ingesting recipes online, then it will not be creating anything 'new'. Innovation and left field ideas might still be the domain of humans.
• and you do still have to know your subject area. I've been using AI at work and by crikey some weird stuff has been served up to by CoPilot. Really important to know your subject, sense check it, figure out what's missing.
• The heart-warming 'human-ness' of the interaction is great. I've used CoPilot to convert a terse email written at the end of a trying day into a much softer, better version and after removing the gushiness and words I would never use, and popping in a bit of casual colloquial kiwi-speak, that email was perfect. Imagine what nefarious sweetheart scammers could get up to with this - truly scary. So, do we start questioning emails and messages wondering if they are real or not? Will this impact our ability to craft well-written emails and messages? 

I'm fascinated by AI and the productivity enhancements I have already realised are fantastic but I'm not blind to the risks and downsides.