Trying to Keep Learning

When we started at LDD, one of the things I implemented early on was starting a LabVIEW User Group. We knew that we needed more people who really knew LabVIEW. So we promoted it (not well) and got a bit of interest.

Some of the people who came to those meetings joined us as the earliest employees. That initial group were amongst the best people I’ve worked with over my career. Part of it was it that this was a group of builders and learners. We were all much younger then, so spending our “spare time” learning the ins and outs of an esoteric programming language was easier.

But an added benefit that I hadn’t realized was the process forced me to practice, to organize my thoughts, to collect my resources and then present them. Just going through that exercise made me MUCH better with the tools. It’s the classic, “Learn one, Do one, Teach one.”

---

The explosion in generative AI in the past year has been astounding. I can’t believe how FAST it’s happening. I’ve always been a crazy optimizer, trying to think of how to do anything faster. How can we sustainably increase the pace using better tools, techniques or processes?

These AI coding tools seem like the perfect fit for that mindset! But I seem to be walking myself in circles because the ecosystem suddenly got SO BIG.

• Is paying for Github Co-Pilot worth it?
• How do you take advantage of ChatGPT when coding?
• What about Code Interpreter/Advanced Data Analysis? How do I use it, and what is it good for? (and why can’t I find it anymore?!)
• Is a custom GPT like Grimoire better?
• Should I be learning the APIs and trying to do agents?
• What about protecting privacy and IP?
• What about using open-source models? How do we run things locally?
• What about non-LLM tools like Physics Informed Neural Networks (hat tip to Nadia for introducing these to me)?

I am feeling an intense pressure like I am falling behind. I mean, this isn’t new for me. I’ve always felt like I was behind where I wanted to be. But this is one of the few times where I’ve felt so flat-footed.

So, what’s the solution for this kind of paralysis? Action. Don’t try to do it all at once. Pick one part, figure it out and then move on to the next most promising thing when it’s done. It might not be the “optimal” tool but getting one tool on the belt is better than constantly trying to figure out which tool to get.

So this brings me to my latest idea. Learn a tool, use a tool, and then teach a tool. I love pair programming with people, but many of my favourite people to dig into this stuff with are BUSY. Trying to schedule something with them seems doomed. But what if it was like open office hours for programming? What if the discussion could be saved and shared again for later?

So why not livestream about these tools? It will definitely provide me with accountability because I detest getting in front of people when I don’t feel “prepared” enough. But even if I’m not prepared, maybe the people who join will show me?

I spend SO MUCH of my time absorbing information, distilling it and then making it actionable. Why am I not sharing it?

So, this year, I will be trying to share a lot more. Even if no one’s on the other end, that’s OK. The process of doing things and putting them out there will be good for me.

So the first live stream will be at 9pm on Thursday, January 18th. I’ve been using Github Copilot a lot, so this seemed like a good place to start. Hope to see you there.