Re: Welcome , Steven to the AI KR (Artificial Intelligence Knowledge Representation) Community Group

Stephen
Thank you for your introduction,
https://lists.w3.org/Archives/Public/public-aikr/2026Apr/0004.html
which I paste below
For some reason, it it did not reach my inbox, apologies for the delayed
response
Please outline your use case and present your problem space, as you suggest
below
so that we may brainstorm accordingly
PDM


Dear all,

I’m Steve Watt, a physician working at the intersection of evidence
generation, implementation science, and AI-mediated knowledge systems, and
founder of People’s Evidence Lab (PEL). I previously spent many years in
Pfizer’s Medical Affairs and evidence-generation organizations, working on
medicinal benefit–risk, real-world evidence, and patient-centered research.

PEL is a small research and consulting group focused on how evidence
travels — or fails to travel — into practice and policy. At PEL, when we
say evidence needs to “travel,” we mean that evidence should retain the
context, provenance, method history, and clarity needed for downstream
users — including humans, agents, and LLM-mediated interfaces — to assess
what to trust and how to act on it. Our concern is not only whether
evidence is rigorous at the point of generation, but whether it remains
understandable, traceable, and decision-useful when it reaches the setting
where action is required.

That concern is what brings me to AI KR. I’m particularly interested in how
existing W3C standards for provenance and credentialing, including PROV-O
and Verifiable Credentials, and related community ontologies such as PAV
could help AI-mediated outputs carry more of their trust-relevant context
with them. In particular, I’m interested in how humans, agents, and LLM
interfaces might recognize — and respond differently to — outputs whose
provenance, methods, or evaluation history are incomplete or opaque.

The strategic topic areas around Privacy and Trust in AI and Agentic
Ontologies feel especially relevant because the same evidence output may
need to travel across multiple downstream decision contexts — clinical,
patient-facing, regulatory, educational, operational, or agent-mediated —
each with different thresholds for trust, action, and accountability, and
each with different constraints on exposing sensitive underlying data. As
AI systems move from generating answers to taking steps on behalf of users,
this becomes not only a communication problem but a
knowledge-representation problem: agents need shared ways to represent
source, method, review status, limitation, intended use, and appropriate
next action.

PEL is working on methods for assessing whether AI-mediated outputs are
trustworthy, usable, context-appropriate, and safe enough for real-world
decision settings — especially when the output is delivered through natural
language and used by patients, clinicians, teachers, or other non-technical
users. I’d be glad to bring that experience into the group’s work where
it’s useful.

I’m new to W3C and the Semantic Web community, so my first goal is to learn
from existing practice. As a concrete first contribution, I’d be happy to
draft a short use-case sketch — for example, on how provenance and
credentialing patterns might help distinguish stronger from weaker
AI-mediated scientific outputs — for the group to react to, if that would
be useful. I’d also be very grateful for pointers to examples, draft
specifications, or related W3C Community Group work where provenance
metadata is explicitly linked to trust or reliability, especially in
agentic AI and natural-language interface contexts.

With best wishes,

Steve

-- 

*Dr Steve J. Watt, MD, MPhil, MRCP*

Founder | People’s Evidence Lab, Inc.

*We make evidence travel — from insight to trusted action*

People <https://www.peoplesevidencelab.com/>’
<https://www.peoplesevidencelab.com/>s Evidence Lab
<https://www.peoplesevidencelab.com/>

stevewatt13@peoplesevidencelab.com | +1 (914) 817-1765 | LinkedIn
<http://www.linkedin.com/in/stephen-j-watt-md>


On Tue, Apr 21, 2026 at 2:18 PM Paola Di Maio <paoladimaio10@gmail.com>
wrote:

> Stephen
>
> Regarding the automated welcome post, it looks like although the
> github file PR was approved, it was committed to the wrong folder
> have been wrestling with dysfunctional github features *or is it me? -
> thanks for informing me it was not working and glad you got the
> link to the welcome wiki page, which is what counts.
>
> It would be great if you could post a brief intro to your org, your
> interest in web standards, with considerations regarding advancing any
> draft spec , or how you d like to engage with other
> CG participants, related initiatives
> or anything in that direction.
>
> The CG needs to encourage participation. especially of companies/org
> entities which may have resources to devote to advancing draft
> specifications,
> as well as anything else that may be of interest of course, provided
> it is related to the scope of work
>
> Please let me know if you have questions and to be reading your posts
>
> Paola DI MAIO
> I
>
> On Mon, Apr 20, 2026 at 8:41 AM Stephen Watt
> <stevewatt13@peoplesevidencelab.com> wrote:
> >
> > Dear Paola,
> >
> >
> > Thank you very much for the welcome, and for sending the orientation
> page. I found it extremely helpful.
> >
> >
> > I do not believe I received the automated onboarding email, and I
> checked junk as well, so I am very glad you shared the link directly.
> >
> >
> > By way of introduction: I am a physician and pharmaceutical medic, and
> founder of People's Evidence Lab. My background is in evidence generation,
> benefit–risk, patient-centred research, and AI-enabled decision support. I
> am very new to W3C and the Semantic Web community, so I am joining very
> much in learning mode.
> >
> >
> > My interest in the group is in how provenance and credentialing models
> might help humans, agents, and LLM interfaces distinguish stronger from
> weaker outputs in scientific and AI-mediated workflows — especially when
> source data, methods, or evaluation history are incomplete, opaque, or do
> not travel with the output itself. I am particularly interested in W3C work
> such as PROV-O and Verifiable Credentials, alongside related ontology work
> such as PAV.
> >
> >
> > If you have any guidance on what is most useful to post first — for
> example a short introduction, a use-case sketch, or a question to the list
> — I would really appreciate it. I would be very happy to post a short note
> to the mailing list.
> >
> >
> > With thanks again for the warm welcome,
> >
> >
> > Steve
> >
> >
> >
> > Dr Steve J. Watt, MD, MPhil, MRCP
> >
> > Founder | People’s Evidence Lab, Inc.
> >
> > We make evidence travel — from insight to trusted action
> >
> > https://www.peoplesevidencelab.com/
> >
> > stevewatt13@peoplesevidencelab.com | +1 (914) 817-1765 |
> www.linkedin.com/in/stephen-j-watt-md
> >
> >
> > ________________________________
> > From: Paola Di Maio <paola.dimaio@gmail.com>
> > Sent: Friday, April 17, 2026 11:19:11 PM
> > To: Stephen Watt <stevewatt13@peoplesevidencelab.com>
> > Subject: Welcome , Steven to the AI KR (Artificial Intelligence
> Knowledge Representation) Community Group
> >
> > Thanks for joining and welcome Steven
> >
> > Plese introduce yourself if you like on the list, and let us know more
> > about your organisation, your interest and plans
> >
> > Here is an orientation page,  please let me know if you have an questions
> > https://www.w3.org/community/aikr/wiki/Welcome_Post
> >
> > An automated welcome/onboarding email should have been sent by the
> > system, can you confirm?
> >
> > Paola Di Maio,
> > Chair
> >
> >
> > On Tue, Apr 14, 2026 at 7:40 PM W3C Accounts Team <noreply@w3.org>
> wrote:
> > >
> > > Dear Chair,
> > > Dear Participant,
> > >
> > > On 2026-04-14 11:40 UTC, Stephen Watt became a participant in the AI
> KR (Artificial Intelligence Knowledge Representation) Community Group,
> representing People’s Evidence Lab.
> > >
> > >
> > > Thank you,
> > >
> > > --
> > > W3C Accounts Team
> > > Please do not reply to this email, it is automatically generated.
> > > Messages sent to this address will not be answered.
> > >
> > > If you need assistance or have questions, please contact sysreq@w3.org
> .
> > >
>