Week 6

Mon 3rd Feb - Sun 9th Feb 2025

Monday 3rd February 2025

AI Ethics WG

Narrative:

As always, we went through the GitHub board starting with items "In Progress" followed by "To Do".

INTERVIEWING

  • Several interview tasks from both Q4 2024 and this Quarter are still pending

  • Several "interview" issues are still open because Vani is checking interview validity - slower process than expected. Perhaps next Quarter we'll train more people to do it too, to speed things up.

  • Several interviews haven't met the validity criteria that we set for payment - WG admin team will agree on thresholds to partially pay interviews that are not quite there.

  • Biggest issue is interviews too short.

  • Validity comprises 2 things - the WG's agreed criteria for payment, and validity for inclusion in BGI Nexus's research corpus - e.g. a too-short interview could still be included in the research.

  • Should we change our criteria for the next round of interviews?

  • Esther will lead a "peer support"-style meeting for interviewers to discuss what we've all learnt.

AGI-24 PAYMENTS

  • People's work at AGI-24 has at last been paid! Thanks to Esther for working with the Foundation to find a way to process the payments. Although we didn't yet get a process in place that can be used in future, we now understand the administrative landscape better, and should be able to establish such a process next time there is direct payment from the Foundation to people in the Ambassador Program.

TRANSCRIPTION

  • Work hasn't started yet - there will be a training session before the end of Feb on how to transcribe, using Intelligent Verbatim style.

  • Several interviewers are happy to transcribe their own interviews, but for those who are not, there might be tasks for other people to pick up.

MATCH FUNDING

  • Esther has been trying to regularise match funding from the Foundation, rather than having to go through a process each Quarter - nothing definite yet

  • Possibly, match funding could support a small funding pipeline for events attendance and similar.

  • We agreed a meeting between Nexus and the WG would be useful to discuss how best to spend any future match funding that is secured from the Foundation to support the work; although the final decision should be made by BGI Nexus

BGI-25 CONFERENCE, ISTANBUL, MAY 2025 We discussed Ambassador Program involvement in this event, in the light of ongoing discussions on event attendance and spending Program reserves. No definite decisions, but we will get Peter's input and discuss BGI-25 again next meeting. We shouldn't leave it too late, to allow time for possible attendees to get visas.

SURVEY-SHARING Survey sharing tasks are now completed - overall cost was $180 ($30 x 6 people). So far, we got 57 complete responses as a result, and no partial responses; and there might be more to come. Might be worth doing more in a future Quarter - if we scale this (500-ish reponses would be meaningful, which would cost maybe $1,500), it is worth it given the value it adds to the research.

Q2 2025 BUDGET We looked at draft budget and discussed briefly - invited comments from members - TBC next meeting

STRATEGY GUILD WORKSHOP Gorga let the group know that later today there will be a Strategy Guild workshop to discuss AI Ethics work; this is part of Strategy Guild's ongoing sessions with different WGs.

Decision Items:

  • We agreed it would be useful to have a meeting between Nexus and the WG at some point to discuss how best to spend any future match funding that is secured from the Foundation; although the final decision should be made by BGI Nexus

    • [rationale] to ensure input from the WG members on their priorities; possibly to create a small funding pipeline to support events attendance and similar, inspired by Clement and Advance's attendance at the ETHiopia event.

    • [opposing] We'd need criteria and an application process

    • [effect] affectsOnlyThisWorkgroup

  • We agreed that survey-sharing work was reasonably successful

    • [rationale] overall cost $180 57 complete responses, perhaps more to come. If we scale this (500-ish reponses would be meaningful, which would cost maybe $1,500), it is worth it given the value it adds to the research.

    • [effect] affectsOnlyThisWorkgroup

Action Items:

  • [action] Vani will finish interview validity checks in next couple of weeks, so that interviews can be paid where eligible [assignee] CallyFromAuron [due] 21 February 2025 [status] todo

  • [action] WG admin team will agree on thresholds to partially pay interviews that are not quite meet validity criteria, especially on length [assignee] CallyFromAuron, LadyTempestt, Sucre n Spice, CollyPride [due] 28 February 2025 [status] todo

  • [action] Vani to set a date for transcription training session by end of Feb [assignee] CallyFromAuron [due] 28 February 2025 [status] todo

  • [action] Vani and Esther to set a date by the end of the Quarrter for a "peer support"-style meeting for interviewers to discuss what we've all learnt. [assignee] EstherG, CallyFromAuron [due] 31 March 2025 [status] todo

Keywords/tags:

  • topics covered: AI Ethics Interviews, interview validity, AGI-24 payments, Interview Transcription, intelligent verbatim, 1P1V, match funding, ETHiopia event, BGI-25 conference, events attendance, Ambassador Program reserves, Strategy Guild workshop, Q2 2025 budget, AI Ethics survey

  • emotions: quick, Only a few of those present spoke, calm, Collaborative

Tuesday 4th February 2025

Governance Workgroup

Narrative:

We agreed the following timeline for the Q2 budget consent process:

🌟 Thurs 6th March - deadline to confirm what the budget caps are for Q2 (in Open Gov meeting?) and the exhange rate to be used for calculating budgets 🌟 Mon 10th March: deadline for WGs to submit Q1 quarterly reports and Q2 budgets. Also - calculate who is a Core Contributor 🌟 Mon 17th March: deadline for Core Contributors to complete consent forms 🌟 Tues 18th March: discuss objections in Gov WG meeting. WGs without any objections are approved 🌟 Sun 23rd March: deadline for WGs with objections to submit a revised budget 🌟 Mon 24th March: 2nd-round consent forms go out, for those WGs that have objections 🌟 Weds 26th March: deadline for Core Contributors to submit 2nd consent forms 🌟 Thurs 27th March: discuss and finalise in Open Gov meeting

We have the dates for Q3 decision (June) and Q4 decision (Sept) worked out on a similar timeline; but we agreed to confirm these nearer the time. We also noted that for Q3, Guillermo/R&D's proposed dashboard will be ready, which might help streamline the decisionmaking process and shorten the timeline.

RE: setting an exchange rate to be used for Q2 budgets - we noted the current low token price, and suggested $0.35, but noted that the final decision doesn't need to be made until early March.

We discussed the suggestions in the "Budget Caps for Q2" document, and decided our approach to the New Ideas budget; new workgroups; and budget cap amounts (see "decisions" below).

In relation to budget caps - we emphasised that any WGs who feel their cap is too low need to prioritise, and there are not many WGs who have outputs that need to be protected regardless. WGs should also be encouraged to use their Q1 quarterly report to explain any negative effects of their budget caps - for example, listing the work that they would have liked to do but have been unable to do because their budget cap was too low - so that the community can assess whether their cap needs to be raised. But we did note that this process is quite slow.

The lobger term suggestions in the "Budget caps for Q2?" doc particularly around the idea of WGs establishing minimums below which they can't operate - will be discussed in a later meeting.

Discussion Points:

  • Dates for Q2 budget consent process

  • What exchange rate should be used for calculating Q2 2025 budgets?

  • Will there be a New Ideas budget for Q2 2025?

  • Will new WGs be permitted for Q2 2025?

  • Q2 2025 budget caps - will they stay the same as Q1?

Decision Items:

  • We agreed the timeline for the Q2 budget decision process. We agreed to set the exchange rate to be used for Q2 budgets in early march. We agreed to confirm the timelines for Q3 and Q4 nearer the time.

    • [effect] mayAffectOtherPeople

  • We confirmed that the New Ideas budget will be reinstated for Q2, and it will be taken from the cumulative remaining amount that is left over from previous Quarters (just over 5k AGIX), rather than taking a percentage from the Q2 budget.

    • [rationale] Because, since there is money that has been previously allocated to New Ideas and not used, it makes sense (especially in a time of low token price) to use it, rather than reducing WGs' budget caps and taking a percentage from the Q2 budget.

    • [effect] mayAffectOtherPeople

  • We agreed that the upper limit that can be spent on New Ideas in Q2 will be 1,500 AGIX

    • [rationale] because even though we have 5k AGIX unspent in total, New Ideas should be used to support small "starter" things rather than high-budget ideas.

    • [effect] mayAffectOtherPeople

  • We agreed that before Q2 begins, we will discuss and agree guidelines on how this budget can (and cannot) be used, basing this on the existing draft doc

    • [rationale] Because we need a process in place for how proposals to use New Ideas budget should be agreed, and guikdelines on what it can and can't be spent on.

    • [effect] mayAffectOtherPeople

  • We agreed that independent new workgroups with their own separate budgets will not be permitted in Q2; but new WGs can be created under the heading of an existing Guild, and costed within that Guild's budget cap.

    • [rationale] Because this will give e.g. regional Guilds the freedom to start new initiatives as they need, within their existing budgets; but it will not require budget caps to be changed to allow budget for completely new WGs.

    • [opposing] Not an opposing argument, but it was suggested that for clarity we shouldn't call these new initiatives "workgroups" as such, but just "neew initiatives.

  • We agreed that budget caps will stay the same (in AGIX) for Q2 as they were for Q1.

    • [rationale] Because core contributors need to see the reports for Q1 which outline any effects of budget caps before being able to decide if they agree that a WG's cap needs to change

    • [opposing] It was noted that this means that if a WG's cap really is too low, they will not be able to change it until Q3 at the earliest, which is quite slow

    • [effect] mayAffectOtherPeople

  • We agreed there needs to be some kind of context-setting or informational material for WGs to explain this need to use their Q1 quarterly reports to explain any issues with their budget caps. This will be discussed in a later meeting.

    • [effect] mayAffectOtherPeople

Keywords/tags:

  • topics covered: budget caps, Q2 2024 budget, consent process, timeline, token price, Exchange rates, Q1 quarterly report, New ideas, Process for agreeing new workgroups

  • emotions: Good natured, low attendance, Discursive, progress made

Wednesday 5th February 2025

Archives Workgroup

Decision Items:

  • We noted that given current low-ish token price, we're not able to start the proposed initiative to create a skilled documentation team.

We'll review this next meeting We might look at creating a video walkthrough for the new version of the summary tool, as a cheaper way to show people how to use it.

  • [rationale] We had this in our Q1 budget: "If token price rises sufficiently above $0.55" (the exchange rate at which Q1 budgets were calculated). Price hasn't risen enough.

  • [effect] affectsOnlyThisWorkgroup

  • Our Quarterly Report draft is open for comments and changes - see https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/231

    • [rationale] Needs to be submitted on 10th March - so we will sign it off in our next meeting on 5th March

    • [effect] affectsOnlyThisWorkgroup

  • We noted that there is a new repo for the new version of the summary tool - see https://github.com/Quality-Assurance-DAO/archive-tool/commits/main/?since=2025-01-01&until=2025-01-31

    • [effect] affectsOnlyThisWorkgroup

  • André, tool development: see https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/240 We noted that André has been working on an Autosave feature for the Summary Tool - he has it locally, and it will soon be rolled out.

    • [effect] mayAffectOtherPeople

  • We discussed the suggestion (from a Treasury meeting last month) of adding a feature to the Summary Tool for data capture for decisions made outside of WG meetings (for example, the quarterly budget decisions). We considered doing this as a new summary template, but decided against this approach, and will keep ideating

    • [rationale] because it might make us too constrained in how we capture these kinds of decisions

    • [effect] mayAffectOtherPeople

  • We noted that Stephen has completed and shared his presentation on philosopher Robert Stalnaker's approach to context, and its relation to AI ethics https://youtu.be/VhYfA0ehw3w?si=N0tc_AW6T3Bx7duU

    • [effect] mayAffectOtherPeople

  • Open source workflow, Stephen: Stephen has been doing some didactic work to document his thinking on what might be a best-practice approach to applying AI to records. See https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/233.

He is working towards a presentation on graph data modelling, and another on how to audit graph data.

  • [effect] mayAffectOtherPeople

  • LLM development: 1st release of graph database in Python is now complete.

Stephen has used a single meeting summary (a Governance meeting from several months ago) as a demonstration of the graph process. He translates the JSON meeting data to a graph structure using Python, then moves it to a Neo4j graph database, and has been working on suggestions for how to query it and verify that it makes sense.

For the rest of February he will be

  • enhancing this so it is easier to use

  • adding additional records to it to see how it handles the connections between different summaries

  • adding an additional layer of data, from the Archives WG GitHub Board https://github.com/orgs/SingularityNET-Archive/projects/1/views/3

    • [opposing] Note that labels are not properly formulated yet - they are taken from the first part of the node name.

    • [effect] mayAffectOtherPeople

  • Ethics policy: see https://github.com/SingularityNET-Archive/SingularityNET-Archive/issues/234 When this is done, we agreed we will share it in AI Ethics WG and the AI Ethics forum on Discord for further input.

    • [effect] mayAffectOtherPeople

  • Tag taxonomy, especially for topic tags: Composability of our data - could a tag taxonomy "sit on top" of our other structured data to help the graph rag process? We agreed that perhaps we do not need to completely dedupe the entire database - but a controlled vocab for tags might still be useful for the future, in the same way as the controlled vocab for names has been. Perhaps create a list that minimally dedupes (stripping out misspellings and similar, and replacing tags like "budget" with "Q4 2024 budget" etc - or perhaps rely on semantic similarity.

We might come back to the insights gained from the tag taxonomy project in future (the types of topics that get discussed, and how they are distributed across different WGs) when looking at the types of questions that can be asked of the corpus.

But the work we thought we needed to do - the exhaustive "tidying up" of the database and correcting / deduplicating tags - is probably not necessary.

Named Entity recognition scripts might relate to this - determining what an entity is (a person, an organisation, a topic?)

  • [rationale] Because the knowledge graph approach we are now using allows for more "messy" data, but we nevertheless need to use proper techniques and be methodical, so people can see how we're doing things. Semantic similarity is of course one approach to finding connections - but for some data elements (e.g. people's names) this doesn't work, as names are abstract and won't display any semantic relation - so this makes a controlled vocab for names more important.

For tags, however, a graph can find sematic similarity between different-but-similar tags - especially the emotional tags - so we might not need to be so assiduous in ensuring that similar tags ("discursive" vs "exploratory" or whatever) are deduped.

  • [effect] affectsOnlyThisWorkgroup

  • Names taxonomy: we agreed we do need to add "All",

    • [rationale] because it is quite often used to asssign action items that everyone in the WG is going to do.

    • [effect] mayAffectOtherPeople

  • We noted that there is a sense in which Read.ai and other generic AI tools erase who we are as a community if they are used uncritically.

They have been designed to understand conventional business and corporate contexts; soi when they ceate meeting summaries for us, they tend to write them as though we are that kind of organisation. Lacking any context about decentralised communities and how they work, AI summaries can subtly miss the point, and can force a record that suggests the Ambassador Program operates like a corporate environment.

  • [effect] mayAffectOtherPeople

  • We provisionally agreed the WG's budget for Q2 2025 https://docs.google.com/spreadsheets/d/1BLCkYgyV0rnt-Ntlqp_DXGCm1t3KieAVMS_HaUamqQI/edit?usp=sharing

Comments will stay open until next meeting, when we will sign it off

  • [effect] affectsOnlyThisWorkgroup

  • Due to low exchange rate ($0.30) for Q2 budgets, we will not be able to afford to continue with archives management tasks of checking, correcting, and chasing meeting summaries.

So rather than push this cost to Q3, as we might have done in the past, we will use it as an opportunity to experiment. In Q2 we will not do any of this work; summaries will be archived "as-is", and any late submissions will be recorded as "no summary given" with no chasing. Then, we can constrain a graph to look only at Q2 summaries, and see what effect, if any, this has on the clarity and accuracy of information we are able to derive, and compare this with previous Quarters.

If we find it has a negative effect, we can look at ways of mitigating this (reinstate management of summaries; progress with setting up a skilled documenters' team; drastically simplify summary tool template and limit the amount of information we can derive; etc)

  • [rationale] To help us asssess if our graph-based process of interrogating the Archives corpus can work well regardless of summary quality. Impetus is partly financial; but partly because we would need to do this type of testing at some point, so it's good to do it now when there is also a financial reason to stop managing the Archives as we have been doing.

Action Items:

  • [action] Vani to approach DeepFunding, to ask about the issue raised in our November meeting about documentation of DF Town Hall: i.e. sometimes timestamps are not great, and additional work needs to be done to document the sessions in a meaningful way - could DF itself could make any further contribution to the work, e.g. by paying someone from the Town Hall team to add the documentation to the summary tool?

and discuss how the Archives dovetails with documentation that DF is doing itself, such as https://community.deepfunding.ai/ [assignee] CallyFromAuron [due] 19 February 2025 [status] todo

  • [action] We noted an issue with tags - people are copy-pasting strings of comma-separated tags into the summary tool, but they are not rendering as separate tags, but as strings. Andre will look into this. [assignee] André [due] 5 March 2025 [status] todo

  • [action] Vani to mention the decision to no longer correct ,meeting summaries in the WG Sync Call at the end of Feb [assignee] CallyFromAuron [due] 27 February 2025 [status] todo

  • [action] vani and Andre to meet to sign off tag taxonomy, and decide whether to implement a controlled vocab for topic tags [assignee] André, CallyFromAuron [due] 5 March 2025 [status] todo

Keywords/tags:

  • topics covered: graph database, Neo4j, summary tool, autosave, decision tracking, Robert Stalnaker, AI ethics, Ai recordkeeping ethics, graph data modelling, Python, JSON, Decentralisation, Decentralization, representing our culture in summaries, tag taxonomy, names taxonomy, controlled vocabularies, comma-separated tags, semantic similarity, recordkeeping ethics, archives ethics, archives management

  • emotions: Discursive, interesting, progressive, wide-ranging, inventive, didactic

Thursday 6th February 2025

Governance Workgroup

Narrative:

There was no agenda for today's meeting. (We had previously agreed that after January, we would try to keep the Thursday Open Governance meetings open, with no predetermined agenda, for anyone to raise a governance issue.)

The meeting was quite small.

So the meeting was more digressive than usual - a gentle conversation touching on a range of topics not directly related to the governance mechanisms of the Ambassador program. We agreed to document in a minimal way, by simply logging the topics that arose.

Discussion Points:

  • Cross-chain collaborations (especially with ETH, given AdvanceAmeyaw and Clement's planned attendance at the ETHiopia Ethereum conference in mid Feb)

  • Cross-chain collaboration as a mindset

  • Adoption of different chains in Africa; which ones are most popular and widely used

  • Clement and AdvanceAmeyaw's plans for engagement at the ETHiopia conference

  • How we keep track of, and follow up on, issues that were raised in the retrospective sessions for previous Quarters' budget decisions

  • How we assess WG outputs and work quality - for example, video view numbers

  • What our priorities are as a Program for WGs' work

  • Is it easier to critique some kinds of WG outputs than others (e.g. outputs that are more quantifiable); and might that lead to some WGs getting harsher critiques than others?

  • How Thursday meetings should go in future - open space, perhaps not recording, perhaps more minimal documentation, discussion of Governance concepts, particularly decentralised governence, rather than making decisions - but with the exception that we might carry over any pending agenda from Tuesdays sessions

  • Tuesdays are the decision-making meeting - Thurs is gentler

  • How to document discursive meetings of this nature - whether simply logging the topics covered would allow the knowledge graph approach that Archives WG is developing to track those topics easily

Decision Items:

  • We agreed that when Clement and AdvanceAmeyaw return from the ETH conference and a retrospective/ feedback session is held, it might be useful to invite Felix.

    • [rationale] Due to his longtime work on cross-chain collaboration.

    • [opposing] none

    • [effect] mayAffectOtherPeople

  • We agreed that the doc "Retrospectives from the past" https://docs.google.com/document/d/1GorEmCJzcrq3b73p1XUa8E7yYD7_UQETt9axt89qff4/edit?usp=sharing should be updated with material from the Q1 budget retrospective

    • [rationale] Because we need to keep track of issues and follow up on them

    • [effect] mayAffectOtherPeople

  • We will document this meeting in quite a minimal way

    • [rationale] as part of Archives WG's current experiments with knowledge graphs, to see if minimal documentation still enables us to track topics as they develop and are discussed in different sessions

    • [effect] affectsOnlyThisWorkgroup

Keywords/tags:

  • topics covered: cross chain collaboration, Ethereum, cross-chain mindset, Thursday vs Tuesday meeting style, retrospectives, open space, WG outputs, assessing outputs, work quality, adoption, Africa, Documentation, minimal documentation, knowledge graphs

  • emotions: Discursive, gentle, wide-ranging, low attendance, calm

PreviousWeek 5NextWeek 7

Last updated