@skirano

–

06 March 2024 5h56

I am actually shocked they didn't see this coming! It's all in the source code. I can't believe this wasn't on purpose.

OpenAI’s Relevance Against DeepMind: A 2018 Strategic Assessment

By

–

06 March 2024 5h53

Here's the text:
CC: Sam Altman , Demis Hassabis Date: Wed, Dec 26, 2018 at 12:07 PM Subject: I feel I should reiterate
My probability assessment of OpenAI being relevant to DeepMind/Google without a dramatic change in execution and resources is 0%. Not 1%. I

Claude 3 Unredacts OpenAI Emails Using Word Length Patterns

By

–

06 March 2024 5h51

I used Claude 3 to unredact this part from the OpenAI emails. What's wild is that they used a "per word" redaction, meaning each redaction length is proportional to the length of the words, so assuming context and word length, this is Claude's guess using the page source:

AI Model Superior Coding Capabilities Drive User Adoption

By

–

06 March 2024 4h24

It's so much better at coding. That's all it takes for me to switch.

Ted Chiang’s Narrative Potential in AI-Related Concepts

By

–

06 March 2024 4h16

You could spin an entire Ted Chiang novel out of this tweet.

AI Model Code Execution Speculation and Output Generation

By

–

06 March 2024 1h07

Yes, it's really wild. I am speculating that it's actually somehow running the code before providing it, because it displays possible outputs a lot. Which may explain that weird disclaimer at the bottom ehere they say "yet."

Claude adds smart progress bars with intelligent output previews

By

–

06 March 2024 1h03

"Hey Claude, could you add a progress bar to my script?" Not only did it intelligently decide what to display there, but it also showed me a preview of what that would look like. It seems to like showing previews of outputs a lot, which shows a deep understanding of the ask.

Claude Outperforms Cursor in Code Debugging Tasks

By

–

05 March 2024 20h32

Yeah, so I had this bug I couldn't figure out, and as you know, Cursor has awareness of the entire codebase and still, it couldn't help. Claude not only fixed the code with a one-shot prompt, meaning copying the code and asking where the bug was, but it even provided suggestions.