Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Chat Notebooks

writings.stephenwolfram.com

Post reply on HN
Print view View original

51–60 of 87 posts

Re: Chat Notebooks

#51

Post by ray__ »

electroly wrote: post #12
DigiDigiorno wrote: post #5

While I think his notebooks and features look pretty useful, he frames the article around having invented "notebooks" 36 years ago before anyone else, and talks about other notebooks not having features that they've had since 1987. This is a pretty odd context to basically describe a new chat feature. Is this a record that should be set straight? I've used Jupyter since it was IPython notebook, but I don't think that…

What's the predecessor with a notebook interface from before 1987? I'm not aware of one. Wolfram loves to claim he invented stuff but this one might actually be true. The Wikipedia article suggests a contemporary in 1987 (MathCAD) but I'm not seeing a Mathematica/Jupyter-style notebook interface before that year. That would seem to have been the year of the notebook interface, and Mathematica 1.0 did come out that ye…

I was always under the impression that Donald Knuth “invented” notebooks in 1984¹. Maybe Mathematica was the first to implement it, but that’s not the impression you get from the article.

¹ https://en.wikipedia.org/wiki/Literate_programming

Re: Chat Notebooks

#52

Post by esjeon »

ray__ wrote: post #51
electroly wrote: post #12

Earlier quoted context omitted.

What's the predecessor with a notebook interface from before 1987? I'm not aware of one. Wolfram loves to claim he invented stuff but this one might actually be true. The Wikipedia article suggests a contemporary in 1987 (MathCAD) but I'm not seeing a Mathematica/Jupyter-style notebook interface before that year. That would seem to have been the year of the notebook interface, and Mathematica 1.0 did come out that ye…

I was always under the impression that Donald Knuth “invented” notebooks in 1984¹. Maybe Mathematica was the first to implement it, but that’s not the impression you get from the article. ¹ https://en.wikipedia.org/wiki/Literate_programming

AFAICT, literate programming does not imply interactive workflow. Also, Mathematica notebooks are extremely flexible thanks to its lisp-like language, and it's not really about writing docs and code. You can draw diagrams, format texts and paragraphs, and embed dynamic/interactive elements. It's indeed the first and the foremost in its direction.

Re: Chat Notebooks

#53

Post by visarga »

SkyMarshal wrote: post #30
connordg wrote: post #23

One of the Wolfram/Chatbook developers here. I'm happy to answer questions or accept feedback about the new functionality. We're very excited about the potential of Wolfram technology + LLMs, and we've got a number of interesting projects underway in this area. Stephen's other recent blog posts linked at the top of the Chat Notebooks post provide a nice tour. The Wolfram/Chatbook[1] package mentioned in the post is f…

What is your sense on how well current AI models can handle mathematics, vs human language? Is hallucinating or inventing info a problem with math too? Do you guys do anything to minimize, mitigate, or solve it? And do you think current AI models will be able to solve as yet unsolved problems in math and physics, perhaps with increasingly large datasets or augments like Trees of Thought or process supervision? Or do…

Not even humans with PhD can do physics research without a lab. How could a model with just a few tokens of context window and no body do that?

Re: Chat Notebooks

#55

Post by Zondartul »

visarga wrote: post #53
SkyMarshal wrote: post #30

Earlier quoted context omitted.

What is your sense on how well current AI models can handle mathematics, vs human language? Is hallucinating or inventing info a problem with math too? Do you guys do anything to minimize, mitigate, or solve it? And do you think current AI models will be able to solve as yet unsolved problems in math and physics, perhaps with increasingly large datasets or augments like Trees of Thought or process supervision? Or do…

Not even humans with PhD can do physics research without a lab. How could a model with just a few tokens of context window and no body do that?

Most of researchers' work is with formulas and reading articles. Even if you had a lab, you'd still need to think long and hard about what exactly you are trying to measure, and then design an experiment on paper before it can be carried out.

So it's possible, in theory, that an AI-in-a-box could do the hardest parts of science aside from experiments. Whether modern LLMs are any good at it is a different question.

Re: Chat Notebooks

#56

Post by esjeon »

DigiDigiorno wrote: post #5

While I think his notebooks and features look pretty useful, he frames the article around having invented "notebooks" 36 years ago before anyone else, and talks about other notebooks not having features that they've had since 1987. This is a pretty odd context to basically describe a new chat feature. Is this a record that should be set straight? I've used Jupyter since it was IPython notebook, but I don't think that…

The notebook interface of Mathematica is simply unique. It is based on a Lisp-like language (Wolfram language), and fully exploits everything that Lisp can offer (and Lisp can get weird infinitely). It provides an unmatched level of freedom and versatility to the end-user. Wolfram is clearly the first one who built a end-user interface this much powerful, and nothing literally comes close, even Jupyter.

So, yeah, seeing that Jupyter is being used more broadly than Mathematica, Wolfram should feel uncomfortable. (Also note that Jupyter is officially inspired by Mathematica.)

Re: Chat Notebooks

#57

Post by esafak »

oslac wrote: post #50

Hard to believe people look at this Chat UX with LLMs as a good interface. Look at the amount of blabbering in the example - why?

Conversational UIs allow users to use a language they already know well.

Re: Chat Notebooks

#58

Post by garfieldandthe »

First sentence he claims he invented sth a long time back, but it definitely has prior art.

I stopped reading right there. If everybody did that then perhaps he'd learn to be able to express things without immediately being full of himself.

Re: Chat Notebooks

#59

Post by garfieldandthe »

meghan_rain wrote: post #42

All I want is a self-hostable open-source GPT4. The community will come up with things that will make the content of this blog post look like hello world todo apps. Can't a wealthy bazillionaire like Elon Musk just drop a few billions and buy the rights to GPT4?

Elon dropped out of openai because of differences regarding direction and conflicts with what Tesla is doing. He then secretly started developing sth similar and asked for a.moratorium on research+dev so his secret thing can catch up.

Re: Chat Notebooks

#60

Post by dgellow »

garfieldandthe wrote: post #58

First sentence he claims he invented sth a long time back, but it definitely has prior art. I stopped reading right there. If everybody did that then perhaps he'd learn to be able to express things without immediately being full of himself.

It’s part of the character. He has a massive ego, but his writing is worth reading.
Post reply on HN