AI Plays Wordle

Alternate Title: Suda trying to prove he's better than AI

I like how Suda still calls ChatGPT "them" instead of "it," like its a person

I think it'd make more sense, if on the first guess you got 6 points, and on the last one just 1, since making the last guess worth 0 means you just have 5 guesses in total instead of 6, and trying to guess the word on the 6th attempt makes it meaningless

Suda:“fester, pester, that’s it.” Me: J E S T E R

6:02 chat gtp guess cat Suda: Nah I heard pen

Super cool 3 letters word that will give you the most hints: "zzz"

In the Chinese name Yao Ming, Yao is the surname, Ming is the given name. Also, Ming is a dynasty in Chinese history. Ming is also a type of Chinese/Japanese font, similar to serif font for western text. Taiwanese people call it Ming, Chinese people call it Song (which is another dynasty in Chinese history), and Japanese people call it Mincho (literally "Ming dynasty").

This was more fun to watch than I expected. And well done for guessing SCREEN, I was thinking ESCHEW. 😉

6:53 You could've guaranteed the solution here, with "-EAR" confirmed, despite there being a million possibilities for the first letter With WYSFGHBN all being possible, you could instead use guesses to eliminate those options, instead of blindly trying to guess the right one With 4 guesses remaining, a guess like "BINS" would obviously be wrong, but would leave "WYFGH". Then you have 3 guesses left, and could throw out "WIFE" to narrow it down to "YGH", with 2 guesses left. Then something like "hype" would eliminate YH, but leave G as the last possible option, and with 1 guess still available! Just throwing it out there as it may be something you find helpful for those kinds of cases in the future~

The little graphics and sound FX editing flourishes you've added make this video 10x more entertaining. Bravo

I love how he says ‘okay huge!’ Then continues to list a dictionaries’ worth😂

Oh my, this was so entertaining! Thank you, recommended <3 You finally suggested something good!

the strategy is that when you got green letters already and there's too many guesses left (like 3 out of 4 and stuff), you can use green spots to try other letters to see if one of them is yellow, then use it in other spots

I’m a PhD student in NLP, and I’m curious about your “dictionary”! If you’re using standard ChatGPT, then it’s not using a dictionary or picking the most common word. It’s picking the word that it thinks a helpful bot is most likely to pick given the earlier conversation.

6:02 Him: give me a random common word that is only 3 letters long Chatgpt: "Cat" Also him: fills in pen

Now suda makes his own ai bot and trains it to basically be him and then he faces him

“Damn I never look at the V” I think that’s for the better Suda

love ur videos man

What I like to do is when there is a huge ambiguity (like with that "*ear") I make an additional guess containing as much ambiguous letters as possible (for *ear i'd guess something like "fine" for f and n; or "pong", to remove (or assure) pear, near and gear)

6:03 Pem? Pen? Wdym it’s cat