ChatGPT

[cd155]

Today, I spent some time playing around with ChatGPT,

Experiment

so I asked: write a program in haskell such as giving a String, and return a String that only has unique characters from it

It returns:

import Data.List (nub)

uniqueChars :: String -> String
uniqueChars = nub

Explanation:

We start by importing the nub function from the Data.List module. nub takes a list and returns a new list with all the duplicate elements removed.

Next, we define a function called uniqueChars that takes a String as its argument.

In the body of the function, we simply apply the nub function to the input String and return the result.

That's it! With this program, you can call the uniqueChars function with any String as its argument and it will return a new String containing only the unique characters from the input.

Some thoughts

It is just very impressed it could give a really good explanation. The overall structure of the ChatGPT seems very complex. The more I use it, the more I feel that is this something we already did in Drasil.

Anything we can learn from ChatGPT to benefit Drasil development?

[smiths]

This is impressive @cd155. I'm not sure how closely Drasil and ChatGPT are related though. ChatGPT uses machine learning. With enough data it can answer a dizzying number of questions. In my opinion, Drasil could never hope to answer as many questions. However, ChatGPT won't always be right in its answers. If we do Drasil correctly, its answers should be correct. In the unfortunate instance that one of our outputs is wrong, we would know how to fix it easily. I don't think the path to fixing incorrect AI outputs is nearly as clear.