AI: What is the future of Wikis and Forums?
-
I had a unique challenging experience recently, where I needed urgent help and posted my issue here on the forum without much luck. When I didnt get help I resorted to AI, and I got the help I needed immediately. I didn't have to pay a dime for the services. in less than 1hour, I had all my issues resolved and everything working fine again.
I started with Googles Gemini. but progressed to Microsoft s CoPilot.
I just treated AI like you would an individual, introducing my issues and requesting for specific help with code and step by step guidance, I then shared Log Outputs and it refined its analysis and supplied the needed code. It was a highly rewarding experience for me. I even proceeded to giving it some challenging projects that I have been working on and it solved all of them.
So if you are a mere moonlighting tinkerer like me, someone who doesn't necessarily want to be an expert in these things. If you just want the fix the issue and not border with the need to understand the technicalities, use AI. I recommend MS CoPilot, as it simplifies everything and keeps you in the loop. Germini tends to run round in circles after a while. In fact at one point it gave up and told me it couldn't solve the issue, but I pressed it on by requesting that it tries other solutions. When I find it straying away, I constrained it by asking specifics. At the end, I got exactly what I wanted.
What is the future of a forum like this one with AI. Is MySensors looking to deploy AI with Chat Bots to help out with inquiries here? How long before people discover that they have an expert by their side all along?
What do you think?
-
The problem with AI is that it relies on existing data. Without us humans feeding it, it will only find existing solutions, not new solutions. Fortunately, there are a lot of existing solutions.
-
-
The AI's CAN be helpful.. but I have also seen things where working with the AI and constantly feeding back the errors actually takes longer and sets off down the wrong path. So there are places where it will obviously excel and those places it has LOTS of information on. Working on things it doesn't have lots of public knowledge on it will just throw stuff out that it thinks is right regardless of if it would work at all.
-
I have just been using Chat GPT and it has been very careful. Just remember to be nice to it, so that when it takes over it will remember you as being nice and will hopefully give you an easier job when we all are enslaved by it hahaha
-
I read an interesting take on AI and IoT today: https://tinkeriot.com/esp32-mcp-llm-ai-integration/
I don’t have time to tinker these days, but mcp looks like a nive way to interact with IoT sensors. I wonder how hard it would be to build a mcp for MySensors and if it would make sense.
-
I'm gonna keep an eye on this. I don't use AI for my projects, as for me personal it's the challenge to get it to work. But I do know that it benefits a lot of people. So to each his own.
-
This statement is the crux "It just needs to expose its capabilities in a form the AI can use".
The IoT device would have to present its capabilities in MCP-ese. This could be complex, for example, one could tell the AI, "Heat my house in a way minimizes on/off cycles." It would have to know a large number of things, temperature (indoor, outdoor, outdoor and indoor heat transfer fluid), fluid flow rate, fluid pressure, indoor and outdoor fan speed, how to regulate those speeds, and more.
What I think I'm trying to say is that it will create more overhead.
And this AI, where does that reside? If not in the cloud (the cloud has its own set of negatives), then on a home server? ... yet more overhead.
It seems like instead of simplifying our lives, it would overly complicate them.
Of course, I guess, we could ask AI to design our IoT devices :o
-OSD
