No. of Recommendations: 0
So, I had a riddle to solve. It actually took me quite a while to come up with the answer (using math that I think is elementary school level, but certainly pre-high school):
Using the numbers 2, 4, 6 and 8; each of them once. Using addition, subtraction, multiplication and division (hint: subtraction not required), create an equation which equals 25
After figuring out the answer, I asked a number of AI's (including ChatGPT, Googles, Microsoft’s, etc.) for their solution.
Every one of them went down a rabbit hole and kept giving equations which they claimed were answers, but where the basic arithmetic was so egregiously wrong that a 5th grader would be embarrassed to be associate with it.
Think of LLMs as English Literature majors.
They are not well-suited for terse languages like arithmetic in this instance.
Much better at verbose "fluffy" languages where the words:concepts ratio is large.
This indirectly answers Jim's question as well.
Use of language is not an indicator of humanness any more than a baby juggling blocks with calculus equations on them understands calculus.
My secret agenda, of course, is to be a humanity zealot who will keep disqualifying AI as human, even as AI keeps muscling into areas that used to be human territory.
