<scribe> Scribe: Ian
ltoth: I am director of standards
of Conexxus
... I attended CES and these observations are about how CES
news affects our industry and retail specifically (notably
convenience)
... first off CES is enormous
... they use multiple convention centers....~180K
attendees
... around 4500 exhibits, plus education, plus specialized
tracks
... AI was a big theme this year.
... I attended an AI session and a "high tech retailing"
session
... the three Ps of AI: Perception, Prediction, Planning
... we heard a lot about ethics in AI
... data bias was one ethics topic
... we also heard from stitch fix...they send you an outfit per
month...they learn what you like over time
... interesting story because they balance AI with some human
interaction...they've found that humans are better in some
scenarios, machines in others
... different weighting factors (e.g., "higher margins for a
given product")
... also a sense that if you are going to make a life-altering
decision, having human oversight should be considered
... John Deere was on a panel..they've been using autonomous
large machinery since the early 2000s
... they said 90% of their equipment is driverless
... they have precise GPS...within a few centimeters of
precision
... Another big topic was 5G
... verizon has ruled out 5G in four cities
... they have a model of
home->street->city->factories->stadia
... if you take a 4G device on 5G there's not much
benefit
... you get real benefit from 5G devices
... Qualcomm and others touting 5G
... Likely mobile phones first, then tablets and other
computers
... but people were talking about 5G all over the show
... for retailers, this could enable a number of new
applications.
... combining AI and 5G:
* personalization and customization (customer recognition, digital signage, AR+VR, customized shelf tags, narrow omni-channel gap)
scribe: if an adult male walks
into the store..and you have an ad up for a prom dress, that
may not do anything, but teen female maybe
... so age + gender might allow on-the-fly digital signage to
target people walking in the shop
... interest in AR (seeing how furniture looks in the
house)
... also interesting to have customized tags (e.g., customized
prices)
... allowing users to use phone (5G) to get more information
about a product on the shelf
... brick and mortar stores are lagging behind online stores in
terms of data and analytics
... but now there are opportunities for in-store learning -
someone who picks up product A might be interested in product
B
Another topic: ROBOTICS
scribe: medical, elder care in
home
... e.g., detect that someone has fallen
... other applications include education (e.g., reading), early
programming skills
... addressing social isolation
... inventory management
... don't want to be promoting something where you don't have
the inventory
... deployment of tech is one thing ("the easy part") but
education and social acceptance are tricky (e.g., nervousness
about robots taking over jobs, etc.)
[We look at some photos]
scribe: first one shows a facial
recognition tool, identifying people by gender + age
... photo shows smart tags with rating, QR code, dynamic
price
... third photo showed a machine to bake bread and to
distribute baked loaves in a vending machine
... you pick a loaf, the machine bags it, and dispenses
it
... another setup shows automation of checkout including "whole
fruit" (which is harder to determine than a package of sliced
fruit)
[photos on smart displays]
scribe: picture of a hat with a
screen, and a handbag with a screen
... these are programmable displays
... some interesting TV displays
... "bezel free, scalable to any dimension"
... LG showed a tunnel of TVs
... saw a lot of 8K TVs
... we saw some screens that are TVs when they are on, and Art
when off.
... instead of a black screen
... there was a lot in the way of pet products (e.g., monitor
pet health, AI to manage feeding...e.g., which pet ate from
which bowl)
... there's also a lot in the realm of games and virtual
reality in gaming
... there was a suitcase that was following a person around
(they did not carry or pull it)
... there was a lot of emphasis on education and learning to
code (e.g., Harry Potter wand, Kano build your own tablet)
dezell: Thank you Linda!
<Zakim> dezell, you wanted to ask about current related W3C work
dezell: What is going on at W3C in some of these areas?
https://www.w3.org/2018/09/immersive-web-wg-charter.html
IJ: Immersive web
... 5G
<manu> IJ: There have been a number of discussions around these themes...
https://www.w3.org/2017/11/web5g-workshop/
<manu> ... Jeff has spoken about Web 5G
<manu> ... There is a whitepaper from Dom on what the Web needs from 5G and how to take advantage of 5G on the Web, that's an interesting piece.
<manu> ... as far as AI in the browser, my gut is to say "AI, that's something that's server side", but it turns out that browsers are becoming platforms that are performing sufficiently quick (via Web Assembly) AI in the browser experiments.
<manu> ... happy to put folks in touch w/ people, both in terms of Data activities, browser capabilities, in looking into these areas.
<Zakim> manu, you wanted to ask about privacy - don't expect many to have actually done research on that.
Manu: Linda, at the beginning of
your deck you talked about image recognition, etc.
... my question has to do with privacy...was there a discussion
about privacy? do customers want to be identified that
way?
... we are working on these sorts of things in VCWG...all of
those technologies at CES seemed to be privacy-eroding
... was there discussion about how people might leave stores
(where there are cameras looking at them) and return to the web
for enhanced privacy.
ltoth: There was definitely
conversation about that in the education track
... especially with regulatory requirements internationally
(e.g., GDPR)
... there were definitely some conversations
... less so in the exhibitions, but I didn't really have time
to speak with tech providers
... but I would say it's top of mind.
Tony: Let me start by saying the official report from the Workshop is not yet available
Tony: December Workshop...I
believe there was a mismatch regarding "what the workshop was
about or should have been about"
... some people came to the workshop wanting to solve the
decentralized ID issue...others wanted to solve the strong
identity and authentication aspects
... I think some people were upset in a variety of
directions
... but I believe that we accomplished trying to understand
what people were thinking about
... we did some voting regarding hot topics
... as far as strong authentication and identity, one theme
that was raised during the meeting related to derived
credentials.
... we seem to be moving towards a strong authentication (on
the Web) we have a legacy of PKI infrastructure in some
jurisdictions
... how does that tie into some of W3C's work regarding
WebAuthn
... how do we link the two? Some jurisdictions have a
requirement that you cannot authenticate unless you've been
identified by the CAT card / PKI card
... so we have that issue to try to resolve
... how to tie PKI work into strong authentication
... I would say that there was little talk about strong
identity, that is proofing aspects --- verification of a
(digital) identity
... there was discussion about claim methodologies, but those
are "after the fact" of proofing process
... there was little on the actual proofing aspects during the
workshop
... I would think that there would be a follow-up to this
workshop (e.g., another workshop, or ongoing discussion on a
list or in a group)
... I think there are a lot of unanswered questions still
around decentralized identifiers
... I think the feedback was mixed; some people wanted to solve
it, others not, or in-between
... I think Manu participated by phone and I welcome his
perspective here as well
<Zakim> manu, you wanted to provide some take aways.
manu: A snow storm presented
travel, so I attended by phone
... tony is right in his assessment that there were different
goals. There was a pretty strong decentralized ID contingent
(looking to get a a WG started; that process is ongoing)
... there seemed to be strong support for the WebAuthn work,
and the potential follow-on work for derived credentials
... we would have liked to have seen more identity proofing
discussions at the workshop
... as an outcome, I would expect the report to be completed
over the next month
... my personal expectation is that there will be follow-on DID
work in the form of a charter discussion
... as Tony said, we still have some hurdles there
... so as far as outcomes, I would think WebAuthn would pick up
some of the topics
... a second would be a potential DID WG (launched in 2019
perhaps)
... and clearly some things not yet ready like DID-based
authentication
... Also some slide decks will be linked from the report
Tony: We saw progress in this
particular area. I think there needs to be more progress at
least as far as strong identity
... I think these issues can be worked through ... the one
lacking is the strong identity piece that needs to be brought
through
... we have deployments of strong authentication (WebAuthn)
<manu> +1 to that.
Tony: we deliberately did not tie
the identity piece into WebAuthn
... so that's the piece that needs the most follow-up
... The Workshop was valuable for hearing where people's minds
were at the time
... there were good contributions and that will be available
from the report
IJ: Is "strong identity" binding between digital identity and real-world identity?
Tony: We can do the strong
authentication piece with crypto
... proof of possession is one thing
... breeder documents is the next thing...some countries have
them, others don't
... "breeder docs" include birth certificates or some form of
legalized proof of who you really are.
... some of the work in the VC claims involve claims and who
made them,
... but whether the claim is true is another question
... for strong identity you need the proof of how the claim was
actually obtained; ISO has done a lot in this particular
area.
IJ: Web browsing is (for many) best anonymized by default. How do you square that with strong identity?
Tony: Some through zero-knowledge
proofs. But you still need some methodology for acquiring the
data.
... you may, for example, believe some passport-issuing
countries over others based on proofing mechanisms.
... you can then put privacy protecting pieces (or anonymity
pieces) on top
<Zakim> manu, you wanted to note that Verifiable Claims WG is working on expressing those things... but not the process... evidence, process, etc.
manu: The VCWG is looking at some
of these processes
... but what Tony said is correct - processes exist e.g., at
ISO and we don't yet have a way to express that in claims
... there are proofs of concept and pilot projects around these
topics, using W3C VC work
... using JWT representations
... the VCWG has put in placeholders for this information in a
VC but we still don't have the data model(s) used to express
this information
... good news is that this stuff is being incubated..I think we
will revisit this later this year or in 2020
By default 25 February, but still to be confirmed
dezell: I have reached out to
some people re: 25 Feb so then we'll let people know shortly
after that
... we also welcome suggestions from the participants here.
Contact Ian or the chairs
... Thanks Tony and Linda (and Manu)