[Joannecoop]

Hey everyone. I am a very keen bridge player and have been playing BBO for about 6 months but am new to this forum. I am trying to develop a Chat GPT-based app to answer very basic bridge and poker questions and would love to hear from you as to what features you think an app like this would need to make it useful. I have made a very quick "mockup". Here is a link to the current very basic version of the app. https://www.joanneco...isewords-mockup
Your feedback is much appreciated.
Joanne Cooper

[pilowsky]

Sounds interesting but your link produces a blank page.

[pescetom]

 pilowsky, on 2023-June-19, 18:46, said:

Sounds interesting but your link produces a blank page.

It works for me.
Maybe you blacklisted .za ?)

[Joannecoop]

 pilowsky, on 2023-June-19, 18:46, said:

Sounds interesting but your link produces a blank page.

Did you manage to access it?

https://www.joanneco...isewords-mockup

[hrothgar]

ChatGPT gets an awful lot of stuff very very wrong.

Using it to generate content for folks who won't understand what's right and what's wrong isn't going to lead anyplace good.

Reposting the following

---------------


I've seen a bunch of interesting claims regarding how well ChatGPT is at generating code.So, for kicks and giggles, I decided to whether I could use it to compare the two different "Best Hand" generation methods that are being discussed on a Bridgewinners.

Before doing so, I decided to test it on a somewhat more simple problem. Here's the question that I posed to OpenAI.

"R" is a high level language for statistical programming. "Bridge" is a card game. Use the R language to create a program that generates 100,000 bridge hands and use this to calculate the mean and standard deviation for the number of High Card Points held.

1. The bad news: The method that it is using is wrong, wrong, wrong
2. The worse news: The results look vaguely plausible

I am trying to decide whether I think that the fact that this produced running code is good or bad


################

Function to calculate the High Card Points in a hand

hcp <- function(hand){ return(sum(pmax(hand - 10,0)))}
# Generate 100,000 bridge hands

hands <- replicate(100000,sample(2:14,13,replace=TRUE))

# Calculate the HCP for each hand

hcp_scores <- apply(hands, 2, hcp)

# Calculate the mean and standard deviation of HCP across all hands
mean_hcp <- mean(hcp_scores)
sd_hcp <- sd(hcp_scores)

# Print the results

cat("Mean HCP:", mean_hcp, "\n")
cat("Standard deviation of HCP:", sd_hcp, "\n")

[hrothgar]

And just tried the following

Al Roth was a very well know bridge player. He was well known as a bidding theorist. Tell me about some of his contributions to the game and about his bidding style.

ChatGPT
I apologize for any confusion, but the information you have provided about Al Roth seems to be incorrect. Al Roth, also known as Alvin E. Roth, is not primarily known as a bridge player or a bidding theorist. Instead, he is a renowned economist and a Nobel laureate in Economic Sciences.

Alvin E. Roth has made significant contributions to the field of market design and matching theory. His research focuses on designing efficient and effective marketplaces, such as the matching of doctors to hospitals, students to schools, or organ donors to recipients. His work has had real-world applications in various domains, including kidney exchange programs and school choice systems.

While Alvin E. Roth's contributions have had a substantial impact on economics and market design, there is no notable connection between him and the game of bridge or bidding theory. It's possible that you may be referring to another individual with a similar name who is more closely associated with bridge. If you can provide more details or clarify any other aspects, I'll be glad to assist you further.

[thepossum]

ChatGPT is helping sort out the wheat from the chaff

It reminds me a bit of travel experiences in different parts of the world. Sometimes incorrect/any advice was regarded as more polite than not helping at all

You can still have a wonderful time

EDIT I have tested one basic question and sadly if I used the advice given I would never get to open the bidding

[pilowsky]

 Joannecoop, on 2023-June-20, 02:34, said:

Did you manage to access it?

https://www.joanneco...isewords-mockup

Unfortunately no - I only get a blank page.
I do see your favicon, the https details and then a blank white page.

[Joannecoop]

So sorry for your trouble. If you want to try again perhaps you can try a different browser? “https://www.joannecooper.co.za/wisewords-mockup”

[Joannecoop]

The idea behind the app is to provide beginners with answers to the basic bridge questions such as “how many points do you myself and my partner combined need for slam”

[Joannecoop]

 thepossum, on 2023-June-20, 18:53, said:

ChatGPT is helping sort out the wheat from the chaff

It reminds me a bit of travel experiences in different parts of the world. Sometimes incorrect/any advice was regarded as more polite than not helping at all

You can still have a wonderful time

EDIT I have tested one basic question and sadly if I used the advice given I would never get to open the bidding

ha! Yes I remember this. Chat gpt will never say it doesn’t know something. It will just take a guess. The tough thing is knowing when it is guessing and when it “knows”. It would be nice if it gave you an indication on his certain it is if it’s answer.