What's new
GrowWeedEasy.com
You're browsing as a guest. Sign up to join the conversation, share your grows, and get help from thousands of fellow growers — it's free and takes less than a minute.

Using AI to grow weed. Will it steer me wrong?

suspencer

suspencer

New member
It seems like AI knows the basics of Cannabis Cultivation. So can you use something like ChatGPT to learn how to grow weed?

I know you should go to a site written by actual human growers like growweedeasy.com and check before trying anything new or crazy to your plants. Like you shouldn't trust AI blindly. Use your common sense. Check pictures of real plants.

But just generally, how good is AI for learning how to grow weed on a scale of 1-10?

What's been grower's experiences using ChatGPT to grow weed? Like how did your plants react. Any horror stories? Any success stories?
 
AI? What’s that . Can you write a paragraph or 2 and fax it to me. Never tried ai. I guess I’m a Luddite ..not to sure about it. I think all the info you need is right here.
 
Last edited:
I know nothing about ChatGpt suspenser, I know what the definition of AI is, but I don't think I have ever used it. There was no computer around when I started growing ganja. Just trial and error for me. I guess I'm locked in that time warp. It's a comfortable spot for me to stay in and that's a win in itself, I think! There are lots of computer savvy people here to answer you. Good luck with your future grows
 
I HAVE used AI and find that its responses are often generalized and frequently WRONG! I have to constantly challenge it to look deeper, turn off algorithms and select which resources it uses and how. I even tested High Times Grow Bot with a specific growing situation . Result was 50/50 at best.

If you are a newer grower, I would forget AI for your grow. The GWE tutorials are far superior! Bottom line: Use GWE to become a master grower.
Use AI to impress your friends.
 
I have used Gemini and ChatGpt. My advice? Don't do it. I gave them both the same questions that I knew the answers to and challenged them when they were wrong. They both guessed a lot. I wouldn't trust AI for growing weed at all. I'm not opposed to using technology for growing. My thought is some tasks and activities simply cannot be accomplished without getting your hands dirty. AI can't do that, at least as we now know it. GWE has the answers. Funny thing is Gemini refers to GWE as the gold standard for home growers.

Right now, I'd give AI a 2. It's good for calculations but since I was being guided to apply nutrients incorrectly I'm skeptical of it overall. If there were a weed specific AI bot I might have a different opinion. Just my .02.
 
Last edited:
Allotrope1000 and golfnrl have the general chatbots right. ChatGPT answers from memory. It has none of your tent's numbers and nothing holding it back, so it guesses. And it sounds just as sure when it's wrong.

I built a custom weed-specific AI system to fix that in my own tent. It runs a local model on a small computer next to the tent, and feeds it live metrics from an array of sensors: temperature, humidity, leaf temperature, light. and performs Vision AI with a camera. Every answer it gives has to cite the published research behind an answer. When it can't, it labels the answer general knowledge instead of passing it off as fact.

It still gets things wrong, just far, far less often than raw ChatGPT. golfnrl, that's the weed-specific bot you were wishing for.

Still check what it tells you, suspencer. Then check it here.
 
Wow gullywompr. Your high-tech world sounds like something that the other high-tech people on this forum may learn from. Although these forum members do stay up on a lot of the latest tools out there. I am just an old guy with old ways and likes it that way, but I can appreciate hearing about the new. Welcome to the forum and hope your high tech set up gives you many ounces of quality ganja!
 
Hey, gullywhomper.
You have a commercial opportunity sitting in front of you. A lot of folks have an old PC sitting around. You could at least write a book with how to put one of your high-tech monitoring systems in place - buy this sensor, hook it up this way, . . . etc.
 
Thanks Allotrope1000, and thanks Chipmunk.

An old PC can do part of it. A small local model will run on most desktops from the last few years. Mine runs on a little board about the size of a deck of cards, with an AI chip built in. It's not fast: about six words a second, so a full answer takes the better part of a minute. But it sits next to the tent all day without running up the power bill.

The sensors are the easy part. You need air temp and humidity, an infrared sensor pointed at a leaf, and a light sensor. Each one runs from a few bucks to about $20.

The hard part is everything after that. The readings have to get into the model. It has to cite the research it's leaning on. Most of mine comes from Youbin Zheng's Handbook of Cannabis Production in Controlled Environments (University of Guelph). And it has to say "I don't know" when the numbers don't support an answer. That's where nearly all my time went, and I'm still working on it.

No book yet, but I've been writing up the build as I go:

- The 3D-printed enclosures, and what went wrong with them: https://croplock.com/blog/the-model-said-it-fit/
- The AI drew the parts, I made them fit: https://croplock.com/blog/the-ai-drew-it-a-human-made-it-fit/
- Sourcing the boards and sensors: https://croplock.com/blog/the-parts-we-cannot-buy/
- Running the model on the AI chip versus the processor: https://croplock.com/blog/the-npu-didnt-make-it-faster/
 

Attachments

  • mars-hotel-summer-26-under-croplock-day-102.jpg
    mars-hotel-summer-26-under-croplock-day-102.jpg
    263 KB · Views: 0
You must log in or register to reply here.
Back
Top