Step 1.
when I tell chatGPT, your #2 solution have fatal flaw,
chatGPT agree, admit #2 is not going to work.( red box )
seems it understood why it not work by the picture I uploaded.
it seems analysis the picture then do some analysis.
Step 2.
chatGPT admit #2 solution use parcel is fallback, admit corner issue.
also suggest go back use MicroStation block range breaker. Where is go back to where I started.
so far none of chatGPT's solution works as it admit.
Step 3.
chatGPT admit its #1 solution isn't work.
Step 4.
chatGPT have better english skill than me,
it understood and explain in better way.
Step 5.
not english native people can actually use chatGPT to write better english.
but the fatal flaw has to be provided by human brain
why chatGPT can't provide these fatal flaw?
I think it because of the foundermental different between AI and human brain.
human brain run slow 10 byte/sec,
but difference is
human brain can truly deeply profoundly understood the problem.
AI quickly search previous people's past experience,
generate solution #1, #2, #3, is mostly rely on that.
Because previous people's past situation is not same as I have today.
chatGPT's solution works on past situation but may not work on my today's situation.
Step 6.
chatGPT does have "brain", it does "think"
see green arrow, green box, it tell me a work around method,
but this method sounds great, but not doable.
if you ask 6 grader, you will get this kind of method.
chatGPT is lacking of profound deep understood the situation.
So it provide solution sounds great, but lack of path to do it.
Step 7.
human brain is not AI, can not search other people's past experience,
human brain can only know his own past experience.
Because of human brain profound understood the situation.
therefor it can filter out the solution that lack of path to do it.
it can filter out the solution that has fatal flaw.
it can generate a solution that is new to the world, never done before by any one.
If you look AI from this angle, AI is NOT , can NOT replacing human brain.
Human brain's #1 solution ( 10 byte/sec ) runs a couple of days
Human brain first filter out these solution has fatal flaw, not doable.
Human brain can only search its own past experience not other people's,
last week, I draw blue line from address point let it intersect with street center line to successfully get street name.
why not continue to draw a orange line from breaker point along the arrow direction ( see orange )
let orange line intersect with previous blue line I draw?
You are not going to see this solution in AI, chatGPT, because no one has done this before including myself.
Step 8.
the fatal flaw is
if I draw same length, red box area each blue line can be intersect by 2 orange range line,
which is not possible to determine use which orange range line.
If I tell chatGPT, your solution have fatal flaw,
chatGPT will admit it, stop on this path, stop going deeper on this path.
chatGPT will go back to original start point pick another "new" path to dig
That is why chatGPT have so many "shallow" path to choose.
But none of them deep enough to see well water.
While human brain don't do that,
facing fatal flaw,
human brain know this path is already "best" compare to those filter out already.
human brain keep digging deeper on fatal path, try to overturn "fatal" to "not fatal".
Step 9.
human brain keep digging on fatal path, try to overturn "fatal" to not-fatal
I can draw multiple time orange line with different length.
I draw long orange line, short orange line.
rule 1). if blue line intersect by 2 orange line, then skip, (can't determin which orange line to use)
rule 2). when blue line intersect by only one orange line, determin range number.
rule 3). repeat draw orange line from shorter to longer, apply rule 1) and 2). until all blue line was processed.
This solution is not going to find on chatGPT, since no one has done before, and this solution need profound understood and deep digging on fatal path.
ChatGPT is not going to deep digging on fatal path, it will choose a "easy" alternative path instead.
This indicate chatGPT lack of deep thinking like human brain.
rule 1,2, 3 requires deep thinking.
Step 10.
summary,
chatGPT create 3 solution, all turn down by human brain.
Because they are "shallow", base on past other people's experience, not applicable to today's situation.
Step 11.
human brain reach a solution by deep digging on fatal path.
which is never done before.
To prove this fatal path is "fatal" or not, is next step.
yellow arrow indicate chatGPT understood what is "profound" understanding the problem.
What is "deep" thinking, why human can determine real-world situation,
Why human keep digging deepper on fatal path.
Why human knows "fatal" is already the best.
These are exactly AI lack of.
Step 12.
Step 13.
Step 14.
Step 15.
Step 16.
Step 17.
Step 18.
Step 19.
Step 20.
Step 21.
Step 22.
Step 23.
Step 24.
Step 25.
Step 26.
Step 27.
Step 28.
Step 29.