LGTM3. Apart from what Daniel and Yoav have written, the non-determinism of this API is going to vary browser version to browser version, as the built-in models are updated. Given that, I believe it's going to be more work than it's worth to discriminate in favor of any single implementation.

On 6/4/25 3:15 PM, Daniel Bratell wrote:

LGTM2

In many ways this is something new for the web. It is a non-deterministic API making it hard to specify or test. It is known to occasionally do the wrong thing. It is currently locked into a very limited language sphere.

Still, everything has to start somewhere, and I think that the API is such that it won't have to change as the implementation improves, making it an API acceptable for officialdom.

/Daniel

On 2025-06-04 19:43, Yoav Weiss (@Shopify) wrote:
LGTM1

While I sympathize with Mozilla's arguments as to the risk here, it seems that no one here is interested in locking-in the web to today's models. I understand the argument that developers may come to rely in weird ways on the outputs of this API as implemented in popular browsers. But that seems difficult to do given the non-determinism of this API's output on any single browser, and the fact that browsers are incentivized to break such reliance (to avoid lock-in to current models).

Given the potential usefulness of this, I think it makes sense to ship the feature, while keeping an open eye for interop issues, in case they'd come.

On Friday, May 23, 2025 at 11:46:53 PM UTC+2 Mike Taylor wrote:

    Thanks - I don't have any further concerns, but it seems useful
    to give Mozilla a few days to respond to the standards position
    ping before approving.

    On 5/23/25 12:53 PM, Daseul Lee wrote:
    Friendly ping!
    Please let me know if there are further questions or comments.
    Thank you!


    On Tue, May 20, 2025 at 9:36 AM Daseul Lee <ds...@chromium.org>
    wrote:

        Please see the inline replies below.

        On Fri, May 16, 2025 at 4:42 PM Mike Taylor
        <miketa...@chromium.org> wrote:


            On 5/16/25 12:14 PM, Daseul Lee wrote:


                    Contact emails

            dome...@chromium.org, kenjibah...@chromium.org,
            m...@chromium.org, btri...@chromium.org,
            ds...@chromium.org, a...@chromium.org


                    Explainer

            
https://github.com/webmachinelearning/writing-assistance-apis/blob/main/README.md
            
<https://github.com/webmachinelearning/writing-assistance-apis/blob/main/README.md>


                    Specification

            
https://webmachinelearning.github.io/writing-assistance-apis/#summarizer-api
            
<https://webmachinelearning.github.io/writing-assistance-apis/#summarizer-api>


                    Summary

            The Summarizer API can convert long input text to more
            concise text, or recapitulate input text with various
            requested options, such as text length, format and/or
            type. Developers may use this API to provide a summary
            of news articles, reviews or blog posts.


                    Blink component

            Blink>AI>Summarization
            
<https://issues.chromium.org/issues?q=customfield1222907:%22Blink%3EAI%3ESummarization%22>


                    TAG review

            https://github.com/w3ctag/design-reviews/issues/991
            <https://github.com/w3ctag/design-reviews/issues/991>


                    TAG review status

            Pending


                    Origin Trial Name

            Summarization API


                    Chromium Trial Name

            AISummarizationAPI


                    Origin Trial documentation link

            https://developer.chrome.com/docs/ai/summarizer-api
            <https://developer.chrome.com/docs/ai/summarizer-api>


                    WebFeature UseCounter name

            kSummarizer_Summarize


                    Risks



                    Interoperability and Compatibility

            This feature, like all built-in AI features, has
            inherent interoperability risks due to the use of AI
            models whose behavior is not fully specified. See some
            general discussion in
            https://www.w3.org/reports/ai-web-impact/#interop
            <https://www.w3.org/reports/ai-web-impact/#interop>.


            By providing a high-level API with clear output
            formats, as well as a capabilities API for detecting
            what a given browser supports, we believe we can guide
            web developers toward using the API in an interoperable
            way that does not depend on the specific models that a
            given browser or browser version uses.


            Gecko: Defer
            (https://github.com/mozilla/standards-positions/issues/1067
            <https://github.com/mozilla/standards-positions/issues/1067>)

            There was a request to comment when these features were
            moving to the I2S stage - can we do to see if Mozilla is
            ready to give a position?

        Yes, we've given the intent to ship status update:
        https://github.com/mozilla/standards-positions/issues/1067
        <https://github.com/mozilla/standards-positions/issues/1067>


            WebKit: No signal
            (https://github.com/WebKit/standards-positions/issues/393
            <https://github.com/WebKit/standards-positions/issues/393>)


            Web developers: Positive
            (https://github.com/WICG/proposals/issues/163
            <https://github.com/WICG/proposals/issues/163>)

            Origin Trial signups have been 5x higher than what we
            see from typical Origin Trials, suggesting a lot of
            developer and partner interest. The survey conducted
            during the Origin Trial, in February 2025, shows
            positive quantitative feedback [1], and third-party
            partners have utilized the API for various use cases
            [2] [3].

            Do you have any qualitative feedback from the OT to
            share, in particular around the API shape or platform
            availability limitations (i.e., has anyone given any
            feedback on the desktop-only nature of these APIs)?

        Platform availability wise, several partners have asked for
        mobile support, which is something we hope to fulfill in
        future iterations, when possible.
        Ergonomics wise, nothing in particular, partners seem to
        like the API (also reflected in the scores we got via the
        readiness survey[1]).


            [1]
            
https://docs.google.com/presentation/d/1DhFC2oB4PRrchavxUY3h9U4w4hrX5DAc5LoMqhn5hnk/mobilepresent?slide=id.g349a9ada368_1_6327
            
<https://docs.google.com/presentation/d/1DhFC2oB4PRrchavxUY3h9U4w4hrX5DAc5LoMqhn5hnk/mobilepresent?slide=id.g349a9ada368_1_6327>

            [2]
            https://developer.chrome.com/blog/summarizer-terra-brightsites
            <https://developer.chrome.com/blog/summarizer-terra-brightsites>

            [3]
            http://developer.chrome.com/blog/summarizer-redbus-miravia
            <http://developer.chrome.com/blog/summarizer-redbus-miravia>


            Other signals:


                    Activation

            This feature would definitely benefit from having
            polyfills, backed by any of: cloud services,
            lazily-loaded client-side models using WebGPU, or the
            web developer's own server. We anticipate seeing an
            ecosystem of such polyfills grow as more developers
            adopt this API.



                    WebView application risks

            Does this intent deprecate or change behavior of
            existing APIs, such that it has potentially high risk
            for Android WebView-based applications?

            None



                    Debuggability

            It is possible that giving DevTools more insight into
            the nondeterministic states of the model, e.g. random
            seeds, could help with debugging. See related
            discussion at
            https://github.com/webmachinelearning/prompt-api/issues/9
            <https://github.com/webmachinelearning/prompt-api/issues/9>.



                    Will this feature be supported on all six Blink
                    platforms (Windows, Mac, Linux, ChromeOS,
                    Android, and Android WebView)?

            No

            Not all platforms will come with a language model. In
            particular, the initial launch focuses on Windows, Mac,
            and Linux.



                    Is this feature fully tested by
                    web-platform-tests
                    
<https://chromium.googlesource.com/chromium/src/+/main/docs/testing/web_platform_tests.md>?

            No

            We have added some web platform test coverage for this
            feature
            
(https://wpt.fyi/results/ai/summarizer?label=experimental&label=master&aligned
            
<https://wpt.fyi/results/ai/summarizer?label=experimental&label=master&aligned>).
            How much we are able to test beyond the surface API is
            unclear, given the nondeterministic nature of the
            output. The explainer discusses this in
            
https://github.com/WICG/writing-assistance-apis/blob/main/README.md#specifications-and-tests
            
<https://github.com/WICG/writing-assistance-apis/blob/main/README.md#specifications-and-tests>.



                    Flag name on about://flags

            summarization-api-for-gemini-nano


                    Finch feature name

            AISummarizationAPI


                    Rollout plan

            Will ship enabled for all users


                    Requires code in //chrome?

            True


                    Tracking bug

            https://issues.chromium.org/issues/351744634
            <https://issues.chromium.org/issues/351744634>


                    Sample links

            https://chrome.dev/web-ai-demos/summarization-api-playground
            <https://chrome.dev/web-ai-demos/summarization-api-playground>

            https://ai.etiennenoel.com/summarizer-api
            <https://ai.etiennenoel.com/summarizer-api>


                    Estimated milestones

            Shipping on desktop

                

            138

            Origin trial desktop first

                

            131

            Origin trial desktop last

                

            136

            Origin trial extension 1 end milestone

                

            139



                    Anticipated spec changes

            None


                    Link to entry on the Chrome Platform Status

            
https://chromestatus.com/feature/5193953788559360?gate=5089800596750336
            
<https://chromestatus.com/feature/5193953788559360?gate=5089800596750336>


                    Links to previous Intent discussions

            Intent to Prototype:
            https://groups.google.com/a/chromium.org/g/blink-dev/c/rCpcU0ZTgTk
            <https://groups.google.com/a/chromium.org/g/blink-dev/c/rCpcU0ZTgTk>

            Intent to Experiment:
            https://groups.google.com/a/chromium.org/g/blink-dev/c/X-VxVeiJgtM
            <https://groups.google.com/a/chromium.org/g/blink-dev/c/X-VxVeiJgtM>

            Intent to Extend Experiment 1:
            https://groups.google.com/a/chromium.org/g/blink-dev/c/u2MRIDCBZQg
            <https://groups.google.com/a/chromium.org/g/blink-dev/c/u2MRIDCBZQg>



            This intent message was generated by Chrome Platform
            Status <https://chromestatus.com/>.


-- You received this message because you are subscribed to
            the Google Groups "blink-dev" group.
            To unsubscribe from this group and stop receiving
            emails from it, send an email to
            blink-dev+unsubscr...@chromium.org
            <mailto:blink-dev+unsubscr...@chromium.org>.
            To view this discussion visit
            
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPhTbOjgsjNHq8YDT5P40P4upEbeXMU-L2N0CaSEWR9z3rhKFw%40mail.gmail.com
            
<https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAPhTbOjgsjNHq8YDT5P40P4upEbeXMU-L2N0CaSEWR9z3rhKFw%40mail.gmail.com?utm_medium=email&utm_source=footer>.
-- You received this message because you are subscribed to
            the Google Groups "blink-dev" group.
            To unsubscribe from this group and stop receiving emails
            from it, send an email to
            blink-dev+unsubscr...@chromium.org
            <mailto:blink-dev+unsubscr...@chromium.org>.
            To view this discussion visit
            
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/597a7785-ae00-4932-8aff-6b1c109ccfbe%40chromium.org
            
<https://groups.google.com/a/chromium.org/d/msgid/blink-dev/597a7785-ae00-4932-8aff-6b1c109ccfbe%40chromium.org?utm_medium=email&utm_source=footer>.

--
You received this message because you are subscribed to the Google Groups "blink-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscr...@chromium.org. To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/9304e1ce-0118-4e48-8fad-49f6fc11ba9cn%40chromium.org <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/9304e1ce-0118-4e48-8fad-49f6fc11ba9cn%40chromium.org?utm_medium=email&utm_source=footer>.

--
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to blink-dev+unsubscr...@chromium.org.
To view this discussion visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/cc2ab776-190a-48ff-a8b4-3fcf0fcbbabb%40chromium.org.

Reply via email to