One negative effect of #LLM s on learning/teaching I have observed multiple times so far:
The LLM sirens leads students towards wrong solutions and I have to be constantly vigilant and guide them back to safety.
Most recent instance:
Discussion
Loading...
Post
A group of master's students had a problem with serial communication between #ESP8266 modules. They asked #ChatGPT for help.
The wise LLM argued that serial communication can only run at 9600 bps (very wrong) which would be too slow for their use case. ChatGPT then suggested that the students should switch to I2C instead - which would have cost them a lot of time and would not have worked well for their use case.
I only noticed this by chance because one of them shared the LLM response in their Discord server (I ask all groups in my courses to set up a server for communication and to invite me).
This was not the first time, and won't be the last time.
:(
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled