In this video, I will tell you about ChatGPT for those who are eager to learn, especially students and teachers.And we're going to do some real scientific simulations together.So I hope it will be different from those videos like ordinary “7 Things You Can Do With ChatGPT” type of videos.
So please watch carefully.
Google Collab code:
colab.research.google.com/dri...
Full text and links: barisozcan.com
Build the Foundation
Let's say we toss a coin.This is a true coin and I’ll toss it for 3 times.How many times do you think all three will land tails?Now I will ask ChatGPT and it gives me the answer right away.The probability of getting tails on each toss is ½, or 50%.Therefore, if I do it 3 times, it is ½*½*½ and will still be ⅛.But I'm not quite convinced of that. It seems a little abstract."Can you list all the possibilities for me?" There.Indeed, there are 8 combinations in total, and in only one, all three lands tails.
Design Your Experiment
Now, how can I be sure that this coin is true?Actually, by doing what I just did.I toss the coin all the time and see how many times lands tails.Since the probability is ½, that is 50%, if I toss it 1000 times, it should be heads 500 times.But it is not exactly 500 heads in each 1000 tosses.There is randomness after all. It might be 502, 498, or maybe 460.But if It is only 10 heads in 1000 tosses, then there must be a problem!One might wonder.So if I toss a coin 1000 times, what would be the difference between the odds of 500 landing tails or 460?But beware,Since it's a complicated thing for it to understand. I give examples here.Out of 1000 tosses, 502 times landed tails.498 times landed tails.I note here.Because I want to see the distribution of it afterward.I just said what I wanted.Mind that, I didn't want to narrow down it’s options by giving a clear directive.Because I am wondering what solution it will suggest me.And I pressed enter!What the!This turned into Python code.Well, I don't know coding. Wait wait.I know you don't know either.I don't understand coding but, Fortunately, I don't need to understand.I'm sure you'll be surprised how simple it is.Our new way of doing wonders is copy, paste, run.Typing Colab into Google.This is Google's very simple interface where we can run Python codes.You don't need to install anything on your computer.I copied the code that ChatGPT gave me.I came here and chose create a new notebook.I pasted it and I just press that play button here.It starts to run, the button turns and turns.Gonna wait a little. In the meantime, let me tell you, I am sharing the codes here in the description of the video, if you want, you can access it from there and try it yourself.And tadaaa!Here is the chart I was looking for.It even put a headline “the results of the coin toss experiment” thank you very much.I see that it has a peak in 500, so at most, there are around 500 heads.When we look towards 400 and 600, there is nothing.That is, out of the 1000 experiments that I tossed a coin 1000 times, none landed less than 400 heads or less.It's always around 500.This is exactly the result we expected.That is unless your coin is fake! You may remember that at the beginning I showed you a site where I found the answer, but which seemed complicated.It is described here, to test whether a coin is fake or not. It's complicated-looking math, lots of explanations, code, graphics and stuff. I've never needed it. just taking a quick glance.My Chart look a little different, I noticed that right away.Mine is pointier. Because Its’ shows between 420 and 580 heads. It is always zero outside of these values anyway. But I don't know how to change the chart! Oh right. We don't need to know. Let me ask ChatGPT. “Can you limit the scale between 400 to 600? Of course".Here is the same chart as on the site.I didn't need to know probability or coding to do it.But it's always important that I understand what's going on!Bell curve is also mentioned on this site. The resulting distribution should fit this function. Well, let's ask for it too, let's see if it fits. New code just arrived.Copy, paste, run. This is it! The mathematical model expected in such a coin toss event almost overlaps with my experiments. Did you notice?We are simulating!We perform simulations using only ChatGPT without knowing almost no math, no coding. In this way, we can learn the nature of probability by testing it.We need to figure out ways to incorporate this into our education.If you want, you can simulate the probability of raising items with probability calculations in the games, develop a strategy in the game by calculating profit / loss, or use it as a material in your lessons.Do scientific experiments at home!But be sure to use it! Because it will change your life.
Researched and Written by: Ögetay Kayalı
Organized and Presented by Barış Özcan
Edited by: Alperen Çatak
The full text of the video and the sources used:
barisozcan.com
25 авг 2023
