"As much as it might feel like some folks are already miles ahead, the reality is that we're just starting to figure out what's possible."
I love this outlook and mindset, and I agree completely (not being technical, mind you- just an observer). It seems as though we're around the time of the Jacquard Loom with regard to AI - we have automated a few tasks that seem really impressive, and that's forcing our own Luddite confrontations, etc - but over time, the explosion in new tech that builds off of the old tech will make today's Jacquard Looms look like hand axes.
Even if you're non-technical, hopefully you're finding ways to personalize the technology for yourself! Simon Willison says that now is the best time to learn to program, since the learning curve has been flattened dramatically with the help of LLMs.
Thanks a lot @charlie guo for putting this together.. and a big thanks for giving the confidence that its still not too late in the game.. was very reassuring for people like me who have started working actively on it only recently :)
Hey Charlie, firstly thanks a lot for even asking that question. Means a lot 🙏. And sincere apologies for the delayed response.. as I didn’t check my notifications earlier. Please find my response below.
So the biggest questions/challenges are:
TL;DR:
- Approach to learn LLM development
- List Practice problems for beginners
- Lack of support community (at least for beginners like me)
1. Which approach to take? : Build & learn while building or Learn first & then build?
I started with the first one as I like to get hands dirty but I quickly realised that I was getting overwhelmed with too much new syntax to learn, and also was supperrrrr slow in doing anything as I didn’t have the full picture. So I took a step back and now doing a deepAI short course on Langchain development. But still this question comes- whether am I going too theoretical?
What would be your suggestion?
2. While I was building the apps, I wondered whether its too difficult a problem I am solving at my level (which is super beginner), as I was too slow and going a lot back and forth, which was quite demotivating. Is there any list of LLM application projects sorted by difficulty level? For ex: when I was learning python, there were many sites which sorted python projects from beginner to advanced level. That was super helpful. Wonder whether something like that exists for LLM projects as well?
3. Support community for LLM development similar to stackoverflow for coding. I tried langchain discord server but didn’t get much support there. This makes getting quick answers to questions difficult. Also I don’t yet have ChatGPT plus subscription, so can’t even use chatgpt for answering those questions, as free chatgpt doesn’t know what is langchain 😂. Any suggestions you have on this one?
Having said all the above, I am overall still enjoying learning this stuff.. and I am sure someday I will be able to build some applications of my own using LLM 💪
"As much as it might feel like some folks are already miles ahead, the reality is that we're just starting to figure out what's possible."
I love this outlook and mindset, and I agree completely (not being technical, mind you- just an observer). It seems as though we're around the time of the Jacquard Loom with regard to AI - we have automated a few tasks that seem really impressive, and that's forcing our own Luddite confrontations, etc - but over time, the explosion in new tech that builds off of the old tech will make today's Jacquard Looms look like hand axes.
Even if you're non-technical, hopefully you're finding ways to personalize the technology for yourself! Simon Willison says that now is the best time to learn to program, since the learning curve has been flattened dramatically with the help of LLMs.
Glad to know I am not super late 😅.. I have recently learned python and now learning how to build using langchain by building mini apps .
I am definitely miles behind anyone in this field but I am still loving it as I am learning a lot of new stuff 😁
Oddly, I feel like I really get prompt engineering. Maybe it's not all that odd, though: it's all just language, after all!
Thanks a lot @charlie guo for putting this together.. and a big thanks for giving the confidence that its still not too late in the game.. was very reassuring for people like me who have started working actively on it only recently :)
You're very welcome! What are your biggest questions and challenges as you're learning this stuff?
Hey Charlie, firstly thanks a lot for even asking that question. Means a lot 🙏. And sincere apologies for the delayed response.. as I didn’t check my notifications earlier. Please find my response below.
So the biggest questions/challenges are:
TL;DR:
- Approach to learn LLM development
- List Practice problems for beginners
- Lack of support community (at least for beginners like me)
1. Which approach to take? : Build & learn while building or Learn first & then build?
I started with the first one as I like to get hands dirty but I quickly realised that I was getting overwhelmed with too much new syntax to learn, and also was supperrrrr slow in doing anything as I didn’t have the full picture. So I took a step back and now doing a deepAI short course on Langchain development. But still this question comes- whether am I going too theoretical?
What would be your suggestion?
2. While I was building the apps, I wondered whether its too difficult a problem I am solving at my level (which is super beginner), as I was too slow and going a lot back and forth, which was quite demotivating. Is there any list of LLM application projects sorted by difficulty level? For ex: when I was learning python, there were many sites which sorted python projects from beginner to advanced level. That was super helpful. Wonder whether something like that exists for LLM projects as well?
3. Support community for LLM development similar to stackoverflow for coding. I tried langchain discord server but didn’t get much support there. This makes getting quick answers to questions difficult. Also I don’t yet have ChatGPT plus subscription, so can’t even use chatgpt for answering those questions, as free chatgpt doesn’t know what is langchain 😂. Any suggestions you have on this one?
Having said all the above, I am overall still enjoying learning this stuff.. and I am sure someday I will be able to build some applications of my own using LLM 💪