3 ms·

This is a very interesting policy. Where to draw the line, though? I.e. how advanced can an auto-complete be in order it to be considered "AI"?

Topfi 2y ago
They point to copyright/licensing concerns, which I do understand considering the way datasets for most currently popular LLMs have been sourced, i.e. large-scale crawling of any accessible source. In that light, most smart code completion offerings should be fine since they historically did not rely on as broad and widely crawled a data set and were generally offered with clear licenses attached. Also, the number of lines these provided historically was lower, which appears to be part of their concern. In the same light, I feel that LLMs that have been solely and verifiably trained on M̶I̶T̶ MIT-0 or 0BSD should be fine, and this may just be a temporary decision until those become more commonplace. In general, though, this will be hard to enforce, but I do understand why they took this stance considering the current flock of LLMs and their training data. Drawing a line in the sand on licensing, even if they cannot truly ensure this with absolute certainty, is understandable and feels appropriate for them, especially considering LLMs have now been in common use for coding tasks for multiple years, so this does not seem to be a knee-jerk reaction.
- pseudalopex 2y ago
  Good points mostly. Why did you say number of lines appeared to be part of their concern? The MIT license requires attribution. You meant MIT-0 maybe? The old requirements relied on people being honest also.
  - Topfi 2y ago
    Thanks. The number of lines aspect goes back to what GNU considers "legally significant" [0]. They are of course not the be-all and end-all in this regard, but felt like a decent reference point when making the post. > You meant MIT-0 maybe? Absolutely, my mistake. [0] https://www.gnu.org/prep/maintain/html_node/Legally-Significant.html https://www.gnu.org/prep/maintain/html_node/Legally-Signific...
femto 2y ago
Given that it's driven by copyright, the limit could be taken to be 15 lines, in line with the FSF's copyright guidance. https://www.gnu.org/prep/maintain/html_node/Legally-Significant.html https://www.gnu.org/prep/maintain/html_node/Legally-Signific...
yjftsjthsd-h 2y ago
The post says > code generated by a large language model or similar technology (e.g. ChatGPT, GitHub Copilot) so I suspect it's actually pretty unambiguous; I've never seen an LLM that was plausibly mistakable for any other technology. (Though if this is merely my ignorance, please suggest an example)

Topfi 2y ago

They point to copyright/licensing concerns, which I do understand considering the way datasets for most currently popular LLMs have been sourced, i.e. large-scale crawling of any accessible source. In that light, most smart code completion offerings should be fine since they historically did not rely on as broad and widely crawled a data set and were generally offered with clear licenses attached. Also, the number of lines these provided historically was lower, which appears to be part of their concern. In the same light, I feel that LLMs that have been solely and verifiably trained on M̶I̶T̶ MIT-0 or 0BSD should be fine, and this may just be a temporary decision until those become more commonplace. In general, though, this will be hard to enforce, but I do understand why they took this stance considering the current flock of LLMs and their training data. Drawing a line in the sand on licensing, even if they cannot truly ensure this with absolute certainty, is understandable and feels appropriate for them, especially considering LLMs have now been in common use for coding tasks for multiple years, so this does not seem to be a knee-jerk reaction.

pseudalopex 2y ago
Good points mostly. Why did you say number of lines appeared to be part of their concern? The MIT license requires attribution. You meant MIT-0 maybe? The old requirements relied on people being honest also.
- Topfi 2y ago
  Thanks. The number of lines aspect goes back to what GNU considers "legally significant" [0]. They are of course not the be-all and end-all in this regard, but felt like a decent reference point when making the post. > You meant MIT-0 maybe? Absolutely, my mistake. [0] https://www.gnu.org/prep/maintain/html_node/Legally-Significant.html https://www.gnu.org/prep/maintain/html_node/Legally-Signific...

femto 2y ago

Given that it's driven by copyright, the limit could be taken to be 15 lines, in line with the FSF's copyright guidance. https://www.gnu.org/prep/maintain/html_node/Legally-Significant.html https://www.gnu.org/prep/maintain/html_node/Legally-Signific...

yjftsjthsd-h 2y ago

The post says > code generated by a large language model or similar technology (e.g. ChatGPT, GitHub Copilot) so I suspect it's actually pretty unambiguous; I've never seen an LLM that was plausibly mistakable for any other technology. (Though if this is merely my ignorance, please suggest an example)