The standard ruby I get from chatgpt and other LLMs is often older ruby syntax. This little system prompt preamble seems to help:

> When giving ruby code snippets, use modern ruby 3.3 syntax, especially for hashes and method signatures with named parameters

Test existing test conversations and production prompts against any new shiny models that appear each day

I will have some capacity for Rails work soon, my DMs are open and retweets are appreciated for reach.
More details:

My ideal setup would be a part-time position(2 days a week) with flexibility of shifting those 2 days and be available 4 days on some weeks and be away for a

In Web/LLM apps, where do you store your system prompts?

Is there a pair of names for using an AI willingly/collaboratively (chatgpt) versus an AI being put in your way (call centres, help windows)?

I think of the latter as “adversarial AI” — you want to get something done and the vendor puts AI in the way.

/cc Simon Willison Ethan Mollick

We hope you enjoy this well researched narrative about Groq, which one of our Groqsters stumbled across. Thank you youtube.com/@Synapse.Offic…
hubs.la/Q02wqnrq0

We're looking for an intern to join us GoRails this summer.

You'll build, test & deploy an open source @Rails app with us. It'll be a great way to get some hands-on, real world experience with Rails and mentorship.

Apply here: forms.gle/JVdGHhUMTZYiae…

Folks, unfortunately my team has been made redundant (company shutting down). I am now looking for the next step in my journey, and if possible, that includes helping my team of extremely talented engineers too. Feel free to DM me.

'easy sounds great!'

Hey everybody, I'm Kasper, I used to be on the Rails core team, but these days I'm helping small/medium Rails shops with their apps.

I'd love to join your team & help you out. Right now I'm looking for 2 clients at ~10h/week, ~3-6 months?

Let's chat. savvycal.com/kaspth/3b22358…

If you’re still using free chatgpt please stop. It’s the model from 2022. Try groq.com and pick the new llama3-70b model from the drop down. Prob about as good as gpt4 and 10x faster and free at the moment.