For those of us that are in the GIS field but cannot wrap our heads around coding. This is a godsend. I can tweak a code here and there but writing from scratch? Nope. I attempted to use it to figure out steps needed to track telecom plant construction progress and costs on a weekly/daily update schedule based on posted asbuilds. It didn't quite get it right, but it did get me in the right direction.
ChatGPT definitely has made my life easier. I work with GIS daily and agree that for a simple task, like SQL query or data transformation ChatGPT will give you sufficient results. For larger projects it is easier to built them on your own. I feel that ChatGPT is here to stay and knowing how to work with it is as much as important than knowing how MS Excel works :) thanks for the review!
I agree that those who think ChatGPT and AI can replace people aren't thinking right. Way back they said the same thing about the internet as well. It isn't that these tools will replace people, it is that these tools will help people work more better and make life easy. I use ChatGPT to write PyQGIS codes. From the simplest task like loading 53 layers from a folder into QGIS to clipping network assets based on certain criteria. While plugins are great, PyQGIS just does the batch work easily and also saves the layers in folders as you want them. I owe ChatGPT for making my life easy and allowing me to stick to the strict deadlines I often have in my line of work. On a side note, co-pilot I find writes codes better than ChatGPT, and I don't know why. Anyone experienced the same thing?
When working with large language models (LLMs) like GPT-3, it's important to be mindful of their limitations and potential for generating inaccurate "creative" or unintended outputs, or hallucination. If you can, please Control the Temperature Parameter (usually using API). The "temperature" parameter in LLMs controls the randomness of the generated outputs. Higher values (e.g., 0.8) make the output more diverse and potentially more creative, but it may also introduce errors. Lower values (e.g., 0.2) make the output more focused and deterministic. I'm using not only OpenAI ChatGPT, but also Google Bard and Microsoft Copilot (and some of other free services). Try to custom prompt, mix and match, and examine the output. Don't blindly accept the answer. I've seen some of the codes provided by LLM are have inner flow. They knew it, but you need to ask them again that you've encounter that error. They will fix it. So .. they may be not giving you their best answer on your first prompt.
@@MattForrest Ah, the ten thousand lake state. I work in forestry, which is basically science in flannel. Got any recommendations for shirt brands? I heard Pendleton used to be all the rage
