Weekly updates:

1. Defined process for definition and prioritization of AI use-cases with the help of @MRaishWMF (Thank you!)
   • Conduct exploratory conversations with product leaders about their views on short-term AI needs and use-cases ( drafted questions here)
   • Compile answers and revise list of existing AI use-cases accordingly
   • Survey PMs for further prioritization of use-case list (very draft questions here, needs more work)

2. Scheduled 3/4 conversations as first step in this process

Weekly updates:

• Refined process and started conversations with Language and ML leadership. Notes are all over the place and need to be consolidated.

Weekly updates:

• Revised list of AI use-cases after completing the conversations with Product leaders
• Documented lessons learnt from the initial conversations
• Designed and distributed a survey to many Product Managers to get further input on prioritization. Deadline for this step is September 2nd.
• Finalized the criteria for AI use-case prioritization:
  1. Priority signaled: has this use-case mentioned during conversations with Product Leadership or as part of the PM survey as a top-priority use-case?
  2. Model availability: have we identified existing models developed externally that can be applied to this use-case? (see T370135#10069044)
  3. Data Availability: do we have enough labeled data to test and if not, what does it take to put it together e.g. through crowdsourcing / manual evaluation?
  4. Measurability: can we in practice estimate the effectiveness of existing models on the proposed use-cases based on quantitative indicators?

Thanks for the work and update, Miriam.

I know you're already thinking about this and I'll emphasize it here: For "prioritization" you will likely need to consult with the AI strategy (development of which is in-progress in T340693), especially if you want to support models for FY26 annual plan work. I would expect much more stability on that front by the end of September, and we should still check in some time later in August because we will have some more clarity after the current deadline of August 20th and getting more feedback.

Sure, thanks @leila for the reminder :)

• Survey input period has concluded. We received 13/16 responses from Product Managers. Thank you all for the contributions!
• We coded the responses as follows:
  • Question What are the top 3 (or less) use-cases you would like to see us develop in the next 12 months was mapped to column H in the prioritisation spreadsheet as the number of direct votes for each use-case.
  • Questions Is there a use-case that is already on your radar for Q3-Q4 or next FY? and Are there high-priority use-cases that are not currently listed? were free text. We manually analysed responses, and added a vote in column H when a use-case was mention. 4 new use-cases emerged from these answers, which are recorded at the bottom of the spreadsheet.
  • We also counted the mentions of the individual use cases in the Are there any use-cases that you think we should NOT consider? question and recorded those in column J.
• Finally we summed in column K the plus (H-I) and minus (J) votes and ranked use-cases accordingly.
• Results
  • The ranked use-cases mostly match the indications given in the initial conversations with leaders:
  • Edit-check related use-cases such as automatically assigning categories to articles (which can be useful beyond edit checks) and policy violations
  • Structured-tasks and mobile-friendly tasks such as automatic article outlines and worklist generation are also considered high-priority.
  • Automated image tagging and descriptions are also highly ranked.
  • Anything that is too close to "automatic free-text generation" looks like it won't be prioritized for now.