5 ms·

DreamCoder: Growing Generalizable, Interpretable Knowledge

83 points by fbodz 3y ago

nimski 3y ago
Building hierarchical abstractions on top of code is IMO the only way to truly enable AI to write code beyond better autocomplete. Kevin's work with DreamCoder shows that hierarchical abstractions can be built automatically in the code domain. Once these abstractions exist side-by-side with natural language (essentially a code-natural language world model), it'll enable arbitrarily complex code generation from descriptions of the outcomes/results.
- remram 3y ago
  Doesn't the existence of this nice, encapsulated, hierarchical code make AI coding useless, rather than easy? You can just use those existing abstractions. The reason AI coding systems exist is because it's hard, with our current programming language and libraries, to do things even if they've been done thousands of time before. If you can build new languages or new libraries/corpus where that's not a problem, you no longer need AI.
  - mirekrusin 3y ago
    It's AI which creates those abstrations itself.
  - Nevermark 3y ago
    If coding is made easier for human’s and AI, by better abstractions, how would that make humans more useful and AI’s less useful? The opposite is more likely. The simpler a task, the easier it is to automate it. The simpler a task, the greater the competitive need to automate it.
iandanforth 3y ago
(2020)
yoouareperfect 3y ago
How's the wake-sleep pattern different from the exploration/exploitation in Deep Reinforcement Learning?
mdaniel 3y ago
Sadness, the "Supplementary materials" url <https://web.mit.edu/ellisk/www/dreamcodersupplement.pdf https://web.mit.edu/ellisk/www/dreamcodersupplement.pdf> is 404 as well as the 2 wayback captures of it are also 404 While trying to track it down, I did find a seemingly related set of GH repos https://github.com/ellisk42/ec/blob/master/dreamcoder/dreamcoder.py https://github.com/ellisk42/ec/blob/master/dreamcoder/dreamc... https://github.com/ellisk42/ecPaper/blob/master/talk.pdf https://github.com/ellisk42/ecPaper/blob/master/talk.pdf

83 points by fbodz 3y ago

Building hierarchical abstractions on top of code is IMO the only way to truly enable AI to write code beyond better autocomplete. Kevin's work with DreamCoder shows that hierarchical abstractions can be built automatically in the code domain. Once these abstractions exist side-by-side with natural language (essentially a code-natural language world model), it'll enable arbitrarily complex code generation from descriptions of the outcomes/results.

remram 3y ago
Doesn't the existence of this nice, encapsulated, hierarchical code make AI coding useless, rather than easy? You can just use those existing abstractions. The reason AI coding systems exist is because it's hard, with our current programming language and libraries, to do things even if they've been done thousands of time before. If you can build new languages or new libraries/corpus where that's not a problem, you no longer need AI.
- mirekrusin 3y ago
  It's AI which creates those abstrations itself.
- Nevermark 3y ago
  If coding is made easier for human’s and AI, by better abstractions, how would that make humans more useful and AI’s less useful? The opposite is more likely. The simpler a task, the easier it is to automate it. The simpler a task, the greater the competitive need to automate it.

iandanforth 3y ago

(2020)

yoouareperfect 3y ago

How's the wake-sleep pattern different from the exploration/exploitation in Deep Reinforcement Learning?

mdaniel 3y ago

Sadness, the "Supplementary materials" url <https://web.mit.edu/ellisk/www/dreamcodersupplement.pdf https://web.mit.edu/ellisk/www/dreamcodersupplement.pdf> is 404 as well as the 2 wayback captures of it are also 404 While trying to track it down, I did find a seemingly related set of GH repos https://github.com/ellisk42/ec/blob/master/dreamcoder/dreamcoder.py https://github.com/ellisk42/ec/blob/master/dreamcoder/dreamc... https://github.com/ellisk42/ecPaper/blob/master/talk.pdf https://github.com/ellisk42/ecPaper/blob/master/talk.pdf