Right? Who cares about flags?

I heard Francis Scott Key had a missing last verse that said, “Lol. JK. Flags are for losers.”

Take autism out of it for a moment and just consider the abstracted communication stream.

Person A has an idea that they want to communicate to Person B. They create a symbol string (ie words) of length N that they believe are isomorphic to that idea.
Person B believes that they have they have captured the entire isomorphism after <N symbols have been received. They start to send their own symbol string in response.

Person A receives this symbol string and decides that it cannot be produced by a process that is isomorphic to their own idea so they say something along the lines of, “No. Let me finish.”

Based on available information, who is more likely to be correct in their assessment?

Do autistic people really believe this?

I generally like to assume that even autistic people have enough self awareness and humility to understand that they don’t know everything.

This thread is full of comments that suggest people don’t actually believe it. Multiple people note that if they actually try to “tune out” they’ll have no idea what the sentence was about when the start paying attention again. That’s direct evidence that the completed sentence was not correctly predicted.

I would challenge anyone who believes this to test it for themselves. Try to predict the actual string of words from the first few words of the other person and see of often you’re correct. There’s no need to tell me the result, just check to see for yourself if you still believe it after the experiment.

If you want to be more rigorous; watch some videos of people speaking candidly, pause the video at random intervals and see if you can predict the next few words.

Thanks. That sounds more like a joke aimed at non-programmers about people who think they’re programmers. If I’m willing to ignore all the valid use cases of AI, I guess it might be funny to assume that none exist.

Can someone explain this joke in programmer terms?

I just wrote an AI that is trained to classify image patterns and automatically creates bitmasks based on the pattern distributions. We tried k-means clustering and got garbage. How would I have done that without AI?

There are a bunch of tools that are basically a text editor hooked up to an LLM. So you use natural language to prompt the software to write code for you.