3 ms·

> But in one sentence, try asking ChatGPT to reverse a string of 20+ digits, like 473936482738338373926. https://chat.openai.com/share/79a7ab39-bd3c-42e8-b4bb-

> But in one sentence, try asking ChatGPT to reverse a string of 20+ digits, like 473936482738338373926. https://chat.openai.com/share/79a7ab39-bd3c-42e8-b4bb-452dedcf8fc8 https://chat.openai.com/share/79a7ab39-bd3c-42e8-b4bb-452ded... > The reversed string of '473936482738338373926' is '629373833837284639374'. So now do you think it understands something?

CamperBob2 3y ago
They'll just move the goalpost some more, as usual. God of the Gaps.
staticman2 3y ago
That only works if you have it run a python script. If you use the non data analysis version it fails.
- insanitybit 3y ago
  And? It knows how to formulate and execute that script. There are pretty hardcoded areas of our brain for doing various tasks - when we offload things to those sections are we not really "thinking" ?
  - staticman2 3y ago
    I'm pointing out the disconnect in the conversation: If you say I'm not smart enough to solve a math problem, and I use a python script to solve it written by a smarter person, whether "I" solved it is a matter of interpretation. By not discussing this I felt the above poster was leaving out some key details.
    - insanitybit 3y ago
      I don't agree with the characterization of the script being written by someone else. The model is writing that script, they're just offloading execution of that script to Python.
      - staticman2 3y ago
        If you object to my terminology replace Python script with "Python interpreter" written by someone else. The point is using someone else's work can be interpreted as not solving the problem yourself.
        insanitybit 3y ago
        Right, the Python interpreter is written by someone else. We agree. But the script is not. ChatGPT understood the algorithm, produced it on its own, and offloaded the execution of that program to CPython. I don't think that that's actually so significant - like I said, parts of our own brains are hardwired to do certain things and the interesting bit is that we know to offload work to those parts.

CamperBob2 3y ago

They'll just move the goalpost some more, as usual. God of the Gaps.

staticman2 3y ago

That only works if you have it run a python script. If you use the non data analysis version it fails.

insanitybit 3y ago
And? It knows how to formulate and execute that script. There are pretty hardcoded areas of our brain for doing various tasks - when we offload things to those sections are we not really "thinking" ?
- staticman2 3y ago
  I'm pointing out the disconnect in the conversation: If you say I'm not smart enough to solve a math problem, and I use a python script to solve it written by a smarter person, whether "I" solved it is a matter of interpretation. By not discussing this I felt the above poster was leaving out some key details.
  - insanitybit 3y ago
    I don't agree with the characterization of the script being written by someone else. The model is writing that script, they're just offloading execution of that script to Python.
    - staticman2 3y ago
      If you object to my terminology replace Python script with "Python interpreter" written by someone else. The point is using someone else's work can be interpreted as not solving the problem yourself.
      - insanitybit 3y ago
        Right, the Python interpreter is written by someone else. We agree. But the script is not. ChatGPT understood the algorithm, produced it on its own, and offloaded the execution of that program to CPython. I don't think that that's actually so significant - like I said, parts of our own brains are hardwired to do certain things and the interesting bit is that we know to offload work to those parts.