Comparisons between LLMs and calculators were made earlier in this thread. That reminded me of this paper https://arxiv.org/abs/2507.19960 which explores these sorts of comparisons. I find that it provides a helpful framework for analysing the relationship between LLMs and cognition. Cheers, Teo
Good paper. Thanks Teo! On Wed, 15 Jul 2026 at 19:57, Teo Camarasu via ghc-devs < ghc-devs@haskell.org> wrote:
Comparisons between LLMs and calculators were made earlier in this thread.
That reminded me of this paper https://arxiv.org/abs/2507.19960 which explores these sorts of comparisons. I find that it provides a helpful framework for analysing the relationship between LLMs and cognition.
Cheers, Teo
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
Having reviewed the updated policy, I think it might be worth reiterating (I think it's already mentioned earlier, but in the introduction, not the policy itself) that use of an LLM to translate documentation to English is acceptable even though it will likely be "substantial" in terms of the percentage of content. It should still be declared, specifying that it was used for translation. On Wed, Jul 15, 2026 at 10:20 AM Moritz Angermann via ghc-devs < ghc-devs@haskell.org> wrote:
Good paper. Thanks Teo!
On Wed, 15 Jul 2026 at 19:57, Teo Camarasu via ghc-devs < ghc-devs@haskell.org> wrote:
Comparisons between LLMs and calculators were made earlier in this thread.
That reminded me of this paper https://arxiv.org/abs/2507.19960 which explores these sorts of comparisons. I find that it provides a helpful framework for analysing the relationship between LLMs and cognition.
Cheers, Teo
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
-- brandon s allbery kf8nh allbery.b@gmail.com
use of an LLM to translate documentation to English is acceptable even though it will likely be "substantial" in terms of the percentage of content. It should still be declared, specifying that it was used for translation.
OK. I added a sentence in P3 Simon On Thu, 16 Jul 2026 at 01:00, Brandon Allbery via ghc-devs < ghc-devs@haskell.org> wrote:
Having reviewed the updated policy, I think it might be worth reiterating (I think it's already mentioned earlier, but in the introduction, not the policy itself) that use of an LLM to translate documentation to English is acceptable even though it will likely be "substantial" in terms of the percentage of content. It should still be declared, specifying that it was used for translation.
On Wed, Jul 15, 2026 at 10:20 AM Moritz Angermann via ghc-devs < ghc-devs@haskell.org> wrote:
Good paper. Thanks Teo!
On Wed, 15 Jul 2026 at 19:57, Teo Camarasu via ghc-devs < ghc-devs@haskell.org> wrote:
Comparisons between LLMs and calculators were made earlier in this thread.
That reminded me of this paper https://arxiv.org/abs/2507.19960 which explores these sorts of comparisons. I find that it provides a helpful framework for analysing the relationship between LLMs and cognition.
Cheers, Teo
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
-- brandon s allbery kf8nh allbery.b@gmail.com _______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
i like the overall shape of the document. i think latent in the llm stuff is actually a sort of value statement in the sortah boy scouts of america sense “be kind, be helpful, and leave work better than you found it” because llms do not have any taste but what you choose to enforce (approx albeit reductive) as long as its about the intent rather than the letter, this looks splendid On Thu, Jul 16, 2026 at 3:41 AM Simon Peyton Jones via ghc-devs < ghc-devs@haskell.org> wrote:
use of an LLM to translate documentation to English is acceptable even
though it will likely be "substantial" in terms of the percentage of content. It should still be declared, specifying that it was used for translation.
OK. I added a sentence in P3
Simon
On Thu, 16 Jul 2026 at 01:00, Brandon Allbery via ghc-devs < ghc-devs@haskell.org> wrote:
Having reviewed the updated policy, I think it might be worth reiterating (I think it's already mentioned earlier, but in the introduction, not the policy itself) that use of an LLM to translate documentation to English is acceptable even though it will likely be "substantial" in terms of the percentage of content. It should still be declared, specifying that it was used for translation.
On Wed, Jul 15, 2026 at 10:20 AM Moritz Angermann via ghc-devs < ghc-devs@haskell.org> wrote:
Good paper. Thanks Teo!
On Wed, 15 Jul 2026 at 19:57, Teo Camarasu via ghc-devs < ghc-devs@haskell.org> wrote:
Comparisons between LLMs and calculators were made earlier in this thread.
That reminded me of this paper https://arxiv.org/abs/2507.19960 which explores these sorts of comparisons. I find that it provides a helpful framework for analysing the relationship between LLMs and cognition.
Cheers, Teo
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
-- brandon s allbery kf8nh allbery.b@gmail.com _______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
_______________________________________________ ghc-devs mailing list -- ghc-devs@haskell.org To unsubscribe send an email to ghc-devs-leave@haskell.org
participants (5)
-
Brandon Allbery -
Carter Schonwald -
Moritz Angermann -
Simon Peyton Jones -
Teo Camarasu