Big Time Bullshit
I have to concede I was impressed for the first few times of using ChatGPT. But then I grew demanding and tested
A) "tell me the difference between the Osprey V22 and the Dornier 31". The result was catastrophically wrong, as it declared the Do 31 being a HTOL aircraft. That would be a 6 in German grading.
B) "write the c++ code for Enigma". The result looked promising at first, until I discovered that the most important ideas were completely lacking. Grade 5.
Given the low complexity of AI systems as compared to "100 billion neurons connected to 10000 other neurons each", I should have known it from the beginning. The human brain is still one of the most powerful information processing systems we know* of. Silicon is on the level of worms at the moment.
In other words, smart teachers will use chatGPT to set up dangerous traps for lazy students.
*Elephant brains can compete on numbers, but I maybe not on structure.
PS: Now the response to the Do31-V22 query seems to be correct. Did they update their model ? Or did I use different wording then ?