Blog post

[Gilad Bracha]

Hello everyone,

Work on Newspeak froze early in the year, when the AI tsunami arrived. Apart from being busy, I had to think about whether the effort made any sense in the brave new world of LLMs.

My conclusion is that yes, there is still very much a point, and my new blog post at 

https://bit.ly/3Lo2MN6

touches upon that. It references, and is best understood, in the context of the prior post, 

https://bit.ly/3iGMp36

written back in January.

The short form is: AI will democratize programming, making it possible for everyone to customize their digital surroundings (and use computers to control physical ones as well).  To fully realize that potential, we will still require live programming systems. Hence, the Ampleforth vision outlined in earlier posts is still relevant.

[Patrick Logan]

Very good explanation of your conclusions and direction. You've provided more detail and more useful information than most of what I've been reading.

I'm well on the skeptical side that a significant step beyond the current "copilot" kind of assistance will be possible with LLMs alone.

I also expect especially logic programming and case-based reasoning to play much larger roles. Your observation of the computational effort required for LLMs alone seems to demand more efficient contributions to the overall solution.

It did occur to me a suitable subtitle to your post might be, "Lambda the Ultimate Symbol Grounding Problem"

[Gilad Bracha]

Hi Patrick,

On Fri, Sep 15, 2023 at 4:20 PM Patrick Logan <patric...@gmail.com> wrote:

Very good explanation of your conclusions and direction. You've provided more detail and more useful information than most of what I've been reading.

Thanks for the comments. 

I'm well on the skeptical side that a significant step beyond the current "copilot" kind of assistance will be possible with LLMs alone.

Yes, I drew a distinction between short and long term, because in the short term I don't see computers as reliable programmers in their own right. All of this is subject to the usual caveats about prophecy; this field is moving faster than any technology I've ever seen. Which is why I lean toward the view that the long term is nearer than people realize.

 

--

Cheers, Gilad