Fortunately these days I have a very patient interlocutor named ChatGPT blessed with an infinite tolerance for half-baked questions and a soothing lack of judgmentality.
I know it’s popular to say AI is overhyped or not actually useful, but I think one area where it’s unquestionably a massive improvement is in asking stupid questions. I ask ChatGPT so many questions that I’ve started looking for ways to automatically send me a daily summary of questions and answers, to add to Anki.
I used to post a lot of these types of questions on subreddits that are Ask-related. Ask Philosophy and Ask Historians, for example. Unfortunately the mod teams for many subreddits like this have gone overboard, deleting anything that doesn’t fit their specific idiosyncratic rules.
Needless to say I have almost entirely stopped asking questions on Reddit.
It would be interesting to compare a person trained in an area, asking questions as a student, with answers from LLMs vs an average set of people. Of course that would require scientific funding and someone experienced in the field to set it up.
Good to recognize stupid questions do exist but so does anxiety around asking necessary questions. The engineer I'm mentoring said they'd ask lot of stupid questions. But we're working through them. A lot of them were things I wondered too, like "this current reading doesn't make sense because the upstream supply is saying it's taking less current" but it was a nonlinear buck regulator. I offered to explain more but they just sat with it a minute, I offered again, but they said they were ok. My viewpoint is to give just a hint and then offer more so they can still go through the thought process on their own, which is really what I'm trying to teach.
> Toss a fair coin until the number of heads exceeds the number of tails. This determines a stopping time. What is the probability that this stopping time is even?
Claude Code did pretty well:
> Zero.
>
> The stopping time always falls on an odd toss. Here's why: after >n tosses, heads minus
> tails equals S_n = X₁ + X₂ + ... + X_n where each X_i takes >value +1 or −1. The sum
> of n odd numbers shares parity with n. For S_n = 1 (the first >time heads exceeds
> tails), we need n odd.
>
> No even stopping time exists. The probability equals exactly 0.
>
> Cogitated for 38s
Fortunately these days I have a very patient interlocutor named ChatGPT blessed with an infinite tolerance for half-baked questions and a soothing lack of judgmentality.
I know it’s popular to say AI is overhyped or not actually useful, but I think one area where it’s unquestionably a massive improvement is in asking stupid questions. I ask ChatGPT so many questions that I’ve started looking for ways to automatically send me a daily summary of questions and answers, to add to Anki.
I used to post a lot of these types of questions on subreddits that are Ask-related. Ask Philosophy and Ask Historians, for example. Unfortunately the mod teams for many subreddits like this have gone overboard, deleting anything that doesn’t fit their specific idiosyncratic rules.
Needless to say I have almost entirely stopped asking questions on Reddit.
It would be interesting to compare a person trained in an area, asking questions as a student, with answers from LLMs vs an average set of people. Of course that would require scientific funding and someone experienced in the field to set it up.
Don't 'the machines' get their info from reddit?
Good to recognize stupid questions do exist but so does anxiety around asking necessary questions. The engineer I'm mentoring said they'd ask lot of stupid questions. But we're working through them. A lot of them were things I wondered too, like "this current reading doesn't make sense because the upstream supply is saying it's taking less current" but it was a nonlinear buck regulator. I offered to explain more but they just sat with it a minute, I offered again, but they said they were ok. My viewpoint is to give just a hint and then offer more so they can still go through the thought process on their own, which is really what I'm trying to teach.
https://math.mit.edu/~apost/courses/18.204-2016/18.204_Jon_L... doesn't this paper demonstrate the same thing like a decade before?
> Toss a fair coin until the number of heads exceeds the number of tails. This determines a stopping time. What is the probability that this stopping time is even?
Claude Code did pretty well:
> Zero. > > The stopping time always falls on an odd toss. Here's why: after >n tosses, heads minus > tails equals S_n = X₁ + X₂ + ... + X_n where each X_i takes >value +1 or −1. The sum > of n odd numbers shares parity with n. For S_n = 1 (the first >time heads exceeds > tails), we need n odd. > > No even stopping time exists. The probability equals exactly 0. > > Cogitated for 38s
I wonder if you could use this to evaluate how "truly random" a random number generator is.