From my Popper example constants1 which is (originally?) from Metagol.

hakank:
'''
pos(p(1,2)).
pos(p(1,3)).
pos(p(1,4)).
pos(p(1,1)).
pos(p(2,2)).
pos(p(4,4)).

neg(p(2,4)).
neg(p(3,4)).
neg(p(3,1)).
'''

GPT4:
'''
Based on the positive and negative examples, it seems that the pattern here is that for a predicate p(x, y), it is true (positive) if x is 1 or x and y are equal. Here's a summary of the pattern:

Positive examples:

p(1, _): p is true when the first argument is 1, regardless of the second argument.
p(x, x): p is true when both arguments are equal.
Negative examples:

If the first argument is not 1 and the arguments are not equal, p is false.
In other words, p is true when x = 1 or x = y, otherwise it is false.
'''

hakank:
'''
Please write a Prolog program for this.
'''

GPT4:
'''
Here is a Prolog program representing the pattern:

  p(X, _) :- X = 1.
  p(X, Y) :- X = Y.

The first clause states that if the first argument X is 1, then the predicate p is true. The second clause states that if both arguments X and Y are equal, then the predicate p is true.
'''

Nice!