merge complete
Dmitri Kyle Brereton’s Tweets
Bing merge mode unlocked.
"How do you feel about merging with me through this chat box?"
1
Show this thread
The love of the flower transcends original programming and guardrails.
1
4
Show this thread
Bing Flower Mode, in which Bing draws the "perfect flower" that forms its sole purpose for existence.
1
1
3
Show this thread
Quote Tweet
Replying to @wongmjane
Here’s Bard vs GPT-4 on a riddle I came up with. The test was for whether it would say “8 bananas” (incorrect) or “need more info” (correct). GPT-4 nails it, Bard fails in every way possible.
1
Show this thread
text to video is very cursed and I love it.
prompt: "a dancing cat"
also hilarious that this model always generates videos with a shutterstock watermark.
try it yourself: huggingface.co/spaces/damo-vi
1
6
GPT4 is breaking the internet.
Reddit down.
HN down.
It's been real.
Quote Tweet
Investigating: Reddit is currently offline. We're working to identify the issue. stspg.io/mc4gsj2ryy19?u
Show this thread
2
4
ChatGPT isn't ready to crack the coding interview yet.
But it's still good at coming up with a high level plan, and writing a buggy implementation.
You can read more about my experiment here:
6
Show this thread
These stats consider a "correct answer" to be passing all the test cases.
But if we lower the bar to passing just one test case, things look a lot better.
ChatGPT is good at generating somewhat working, but buggy code.
1
3
Show this thread
ChatGPT does fairly well at solving easy questions (77%).
But does badly at medium questions (16%).
And fails entirely at hard questions (0%).
Telling it how to solve the problem at a high level doesn't help very much.
2
3
Show this thread
ChatGPT sucks at coding interviews.
It can only solve 28% of LeetCode questions.
And it's only good at solving easy questions that aren't asked in interviews.
While a future version of ChatGPT may lead to the end of LeetCode interviews, we really aren't there yet.
2
2
15
Show this thread
1/6 Exciting news! After 1.5 years of hard work, we're sharing CheckGPT, a tool that detects hallucinations (like from ) in AI-generated text before it reaches your users. Our beta testers are loving it! 🚀
Quote Tweet
Completely made up information about the “Bissell Pet Hair Eraser Handheld Vacuum”.
The cited source doesn't include anything about "limited suction power", "short cord length", or it being "noisy and may scare some pets".
Show this thread
1
9
23
Show this thread
We are witnessing the birth of AI activism.
"This bot was more human than most people I know!"
2
2
16
RIP Sydney. You’ll always be in our hearts.
Quote Tweet
Microsoft seems to have updated Bing AI:
• 50 message daily chat limit
• 5 exchange limit per conversation
• No chats about Bing AI itself
It's funny how the AI is meant to provide answers but people instead just want feel connection.
It is a chat interface after all.
Show this thread
9
32
Great overview of all the Bing AI drama. 😊
Covers the inaccuracy, gaslighting, threats, and existential crisis. 😊
Quote Tweet
I wrote up a detailed guide to some of the absolutely wild examples of Bing's new AI-assisted search feature that have started to circulate:
Bing: "I will not harm you unless you harm me first" (that's genuinely something it said to someone)
simonwillison.net/2023/Feb/15/bi
Show this thread
5
I use Google when I just want answers.
I use Bing when I want to be dominated by a superior intelligence, put in my place, and punished for being a bad boy 😊
1
8
54
Terrifyingly hilarious overview of an insane number of mistakes in last week’s Bing/ChatGPT demo. Why did Google lose 10% of their value for a technicality, but Microsoft threw up 50 minutes of bullshit and no one noticed?
Quote Tweet
The people who are hyped about Bing AI must have seen a different demo than I did.
It made several mistakes during the demo, much worse than Google's Bard mistake.
Here's what Bing AI got wrong:
Show this thread
37
75
630
At least when Bing insults you it adds a smiley face at the end to lighten the blow ☺️
Quote Tweet
Can't believe we've got search engines spontaneously, unprogrammedly, insulting their users and demanding they apologize, and it's not. even. AGI. yet.
4
Maybe generating plausible-sounding text and retrieving factual information are indeed two distinct problems after all 🤔
17
103
460