- From: Domenic Denicola <notifications@github.com>
- Date: Thu, 21 Aug 2025 20:45:46 -0700
- To: w3ctag/design-reviews <design-reviews@noreply.github.com>
- Cc: Subscribed <subscribed@noreply.github.com>
Received on Friday, 22 August 2025 03:45:50 UTC
domenic left a comment (w3ctag/design-reviews#991) > We discussed this on our last call and decided to "time out" and follow up on the Prompt API [#1093](https://github.com/w3ctag/design-reviews/issues/1093), as they have very similar properties. Thanks, this seems like a reasonable way for the TAG to consolidate work. Note, however, that there are significant differences between the "raw" prompt API and higher-level task APIs like the writing assistance APIs or the translator and language detector APIs (#948). So we would appreciate it if the TAG carefully called out any concerns that are prompt API-specific, instead of grouping all of them together. (For some discussion of those concerns, see https://github.com/mozilla/standards-positions/issues/1067#issuecomment-2921239314 and following discussion. A top concern is whether high-level task-based APIs can treat the underlying model as an implementation detail. We believe this is definitely possible. Whereas with the prompt API, it's a much trickier question.) -- Reply to this email directly or view it on GitHub: https://github.com/w3ctag/design-reviews/issues/991#issuecomment-3212932088 You are receiving this because you are subscribed to this thread. Message ID: <w3ctag/design-reviews/issues/991/3212932088@github.com>
Received on Friday, 22 August 2025 03:45:50 UTC