December 15, 2025 • 68 mins
What if your glasses could spot a deepfake before your gut does? We sit down with Jean Marc Seigneur, a veteran researcher of decentralized trust, to map where security failed, where it’s catching up, and how proof—not vibes—will anchor the next decade of digital life. From central bank digital currencies to NFTs that carry qualified electronic signatures, we unpack how legal recognition and cryptography can finally meet in the middle, turning tokens into enforceable rights and payments into reliable public infrastructure.
We also go beyond buzzwords to the missing pieces: education and design. Friendly apps hide sharp edges, so we talk about why countries need their own experts, not just imported tech, and how wallets must evolve with safer recovery, better defaults, and interfaces that explain risk without slowing you down. AI raises the stakes, so we explore signed videos, verifiable identities, and provenance trails that help you tell a real voice from a cloned one at a glance. Reputation won’t live on a web page for long; it’s moving into the physical world as augmented overlays that can help or harm depending on what they reveal and to whom.
Bias won’t vanish either, because human trust is social and local. We discuss how to balance peer signals with regulators’ oversight, why transparency about AI use will give way to tracking human effort, and what a time-based “work token” could add to creative markets. The red thread across it all—payments, NFTs, augmented humans, and AI media—is simple and demanding: protect freedom while proving claims. If we want technology that empowers rather than deceives, we have to design, debate, and defend the trust layer itself.
Enjoy the conversation? Subscribe, share with a friend who cares about digital trust, and leave a review to help more curious minds find the show.
Check out "Protection for the Inventive Mind" – available now on Amazon in print and Kindle formats.
The views and opinions expressed (by the host and guest(s)) in this podcast are strictly their own and do not necessarily reflect the official policy or position of the entities with which they may be affiliated. This podcast should in no way be construed as promoting or criticizing any particular government policy, institutional position, private interest or commercial entity. Any content provided is for informational and educational purposes only.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 2 (00:00):
My my my research is about trust.
And so of course when there isa new system uh emerging and
created, I'm going to look at ifis it trustworthy?
How could you hack it or uh howcould you improve uh its
trustworthiness?
Speaker 1 (00:18):
Um welcome to our podcast.
Speaker 2 (00:29):
Thank you for the introduction.
Speaker 1 (00:31):
Um so the first question is please uh let us
know who you are and how um youcame to be this uh IP expert.
Speaker 2 (00:40):
I mean um uh I've been working on uh decentralized
trust since uh 2000.
I started my research atTrinity College Dublin, I did my
PhD there on this topic.
Uh then after I joined theUniversity of Geneva, where I uh
teach uh online reputation aspart of the Media Lab of uh the
Faculty of Um Social UhSciences, and um uh as well as
(01:07):
directing the certificate ofadvanced studies on blockchain
technology, which is also uh atechnical trust mechanism.
Speaker 1 (01:16):
Yeah, perfect.
Um so you have been working ondigital trust since the early
2000s, that's uh quite a longtime.
Um, what has changed the mostin how trust is built online?
Speaker 2 (01:31):
Uh uh so online uh for the last 20 years uh it's
got it's gotten worse uh becauseuh the technology uh hasn't
changed much from a securitypoint of view.
Um and uh the hackers uh haveno uh know better how to attack
(01:54):
those systems where security hasnot been enough uh uh sought
and and uh invested in.
Uh so uh now there are moreattacks and and security is less
uh as before, uh almost asbefore.
Um but in the last years we wehave seen coming up uh stronger
(02:17):
mechanisms, uh maybe linked touh chips on identity cards where
you could really prove your uhreal identity uh as a human uh
citizen of a country uh online,and uh it's going to help
securing more uh the internet.
Uh but so far no, I mean forthe last uh 20 years uh not much
(02:40):
has been done because uh notsecurity has not been taken into
uh consideration as a strategicuh uh asset.
Um even countries uh have noteven invested in in uh digital
uh technologies.
Uh some countries have, like uhthe USA or or China, but uh
(03:00):
Europe, Europe, for example, oreven Switzerland, they are
talking about uh digitalsovereignty, but uh they have
invested nothing in the coretechnologies, uh and so uh it's
very weak now.
Speaker 1 (03:13):
Okay, so the focus has been to digitalize and to
become more modern, but notthinking about the risk of that
uh digitalization of being moreexposed to the risk online.
Speaker 2 (03:25):
Yes, and and we see different risks because they are
like independent hackers butalso uh like aggressive
countries now.
Oh okay.
Um well, and so uh the internetis uh is not uh uh always uh a
playground where we only uhplay.
Speaker 1 (03:43):
Everyone is there, the good and the bad.
Okay, makes sense.
Makes sense.
So you have helped create oneof the first academic programs
on blockchain and decentralizedapps in Geneva.
Why is education a criticalpillar in shaping through tech?
Speaker 2 (04:01):
This is very important because otherwise you
you seem to uh to know how touse the those systems because
they are very no uh they areuser-friendly, uh, but you don't
know the underlying uh risks.
Uh and so um it's important forthat in each country uh a good
number of people uh know uh thedetails uh and know how to
(04:24):
protect uh their citizens.
Speaker 1 (04:26):
So education is key in this sense.
Speaker 2 (04:28):
Yes, if you don't know the details.
I mean uh uh you rely on onother countries who would
provide you uh we will uh whichwill provide you some um like
technologies that they say aresecure, uh but you don't know.
Uh you you need to to do peerreview and also internal peer
review.
Speaker 1 (04:47):
To verify yourself just in case.
Speaker 2 (04:49):
Yes.
Speaker 1 (04:50):
Yeah, makes sense.
So and also to make sure thatyou learned what the technology
behind it and you understand umwhat is what exactly it's about
in order to be able to regulateor to use it properly.
Speaker 2 (05:02):
Yes, especially when you have hackers or aggressive
countries, you know, who providesome technologies.
Speaker 1 (05:07):
Yeah, makes sense.
So from CV CBDCs uh to NFTs,what's the one sample um where
decentralization improved orchallenged public trust?
Speaker 2 (05:22):
This is very broad because CBDC are central bank
digital currencies, uh, whichmeans that uh the central
central bank would uh create uh,for example, in Switzerland a
digital franc or in the US adigital dollar.
Uh it's another way of paying.
Uh now we have coins, uh, wehave uh banknotes.
Um in some countries we havecheck.
(05:44):
Uh then you can pay also withapplication, uh like uh online
payment.
Uh you have cryptocurrencies,and here uh uh the idea would be
that you have uh uh a money uhwhich is uh uh also provided as
another uh with another means asa digital currency, but it
(06:04):
would be like a real dollar oruh real uh Swiss franc.
There are many different waysto do it.
Uh usually uh we're talkingabout cryptocurrencies, like uh
an equivalent of uh it would notBitcoin but uh a dollar that
would be created uh as acryptocurrency by the Fed or uh
(06:28):
in Switzerland the SwissNational Bank.
Uh but it's not the only way todo it because on the blockchain
cryptocurrencies are also on ablockchain and they are tracked,
and from a privacy point ofview, it's not so so good.
So there are other uh ways todo it.
So uh CBDC are centralizedbecause they are uh provided by
(06:49):
uh a central bank of a country.
So it's very different than uhNFTs, which are non-fungible
tokens, uh, which are usuallyused to uh yes uh you you you in
your crypto hardware wallet orcrypto software wallet, you can
have uh cryptocurrencies butalso those uh NFTs, non-fungible
(07:10):
tokens, uh, that can be linkedto uh a certificate of ownership
of uh piece of art or um ofsomething in the real world,
which is um supposed to be umlike uh real ownership.
Okay.
Uh one piece which is missingin current NFTs is that okay, on
(07:33):
the blockchain this is secure,this is uh really in your uh
when it's it it's written in theblockchain uh and you own this
uh NFT in your crypto wallet.
Um yes, you have it by and it'sprotected by maths, by by
cryptography, uh on theblockchain, in your uh crypto
wallet, but lots of judges uh orlaws uh in different countries
(07:57):
they don't know aboutblockchain, so it's very
difficult to explain to them uhyou know uh yes, this is my
ownership because this is my uhcrypto wallet.
Uh so uh it's difficult toprove really that uh you own it
really in the real world.
Um so then we have worked withuh International
Telecommunication Union uh for anew standard called the Sign
(08:18):
NFTs, meaning that with the NFTuh it's inside your crypto
wallet, but also uh you have auh digital signature which is uh
uh of very high uh strength,qualified electronic signature.
And uh then when you open thisuh NFT, you have also a link,
(08:39):
for example, for a PDF, and thisPDF will be signed digitally
by, for example, the artist, uhand it's equivalent legally as
uh uh the handwritten uhsignature um by law in
Switzerland or Europe, and Ithink there are some uh it will
be also recognized in other uhjurisdictions.
(09:02):
So um, yes, I mean so then yousee from CBDC, which is
centralized, to NFTs that can uha piece of art can be created
by any artist in the world.
Uh you have a big differencebecause it's uh centralized and
NFTs are more in a decentralizedway because anybody can create
NFT, uh a central bank currencycan only be created by a central
(09:24):
bank, right?
Um, and so uh the um last pointof your question was like so
it's um what is the thechallenge or the improvement on
public trust?
Speaker 1 (09:36):
From what you're saying, the one with the the the
extra proof for the NFT, that'sgreat for trust because then
you have an authority that isconfirming the validity of of
this um of the of the art or thework behind the NFT.
Um and what the other one aboutthe um the the one that is uh
(09:56):
managed by the banks, you havethe banks, people usually trust
banks in the majority.
Um what challenges do you seein this kind of because uh in
this kind of development?
Because um if the um coin isgonna get all digitalized, um it
can get uh as well, it can getuh lost or it can get uh uh
(10:21):
break down or someone can attackthat you can lose everything
because it's only digital.
Um so in this case, how do youbuild trust on that?
Speaker 2 (10:30):
Yes, alors so um uh from uh what I understand is
that also from the public pointof view, um then all of this is
is very new.
Uh and so uh there's a bigquestion in uh the population or
the citizen trusting uh thesenew cryptocurrencies.
Uh so from uh a CBDC point ofview, there is uh for the normal
(10:56):
citizens who uh who try totrust in this system, if it's
provided by the central bank, uhthen already from a legal point
of view there is no uh trustissue because this is a legal
tender uh um they are regulatedand so you know this is a uh um
a governmental organization.
(11:18):
Um and so then uh uh from thispoint of view, uh there is less
the the trust question is lessimportant because this is uh uh
provided by uh a by law uh aninstitution that uh says this is
uh uh if you have such CBDC inyour crypto wallet, then you you
you own this uh this money andthis is an official money.
(11:41):
Um so here there is a less lessrisk.
Uh from NFTs, then you you youalso have this education for for
uh uh normal people you know inSwitzerland or in developed
countries, you have between 10to 15 percent of the population
who have already bought somecryptocurrencies, uh, but
(12:02):
sometimes via intermediaries.
Um, you can you you don't needyour own crypto wallet, you can
buy it from some banks.
Uh for example.
Uh in Geneva, you you have aArab Bank Switzerland, for
example, uh you have Swisscoat.
Um and uh so even some of thesepeople don't understand really
(12:24):
uh don't have their own cryptowallet, so uh you have a lot of
of uh uh education to do to forthem to really be uh uh used to
use this new technology.
As the web, you know, it tooklike 20 years, for example,
before there was a large scaleadoption.
(12:45):
Uh and then also for the cryptowallets you need uh you need uh
other components for at theuser interface level that will,
as you said, help not to losethe the the keys for the uh the
crypto wallets, uh to recoverit.
Uh and um for the next fiveyears uh there are some
(13:10):
different working groups workingon uh bringing um easier user
interface for cryptocurrenciesand NFTs.
And uh already, as I said, NFTsuh they miss uh already uh an
official digital signature ofthe artist, or you know, like
the the company that would uhsell um um let's say a flat uh
(13:36):
or uh house uh ownershipcertificate.
Uh you need to add somequalified electronic signatures.
But the good thing is that uhnow in Europe and Ireland you
have these.
This is not part of blockchain,but is this is another part
technology technological part.
Uh most you uh most uh nationalIDs will have chips uh uh uh
(14:01):
that will be uh certified by anumber of trusted parties.
Uh in in Switzerland, forexample, for the arts, uh
Swisscom is uh is one of thesetrusted third parties.
Um as I said at the beginning,now we we are moving into uh uh
maybe an era where it's a biteasier to be sure who uh uh who
(14:26):
is talking in front of you, evenif you have all these AIs that
you could impersonate uh theface, the image, and so on, they
will not be able to impersonatethe mats and the signatures.
Uh so um uh yeah, at themoment, even with AI, it's even
uh more difficult because ifhackers are uh building AI of
(14:47):
people uh with a voice, theface, yeah, all the information
which are uh known about us.
Uh this interview, for example,uh already uh an AI could be uh
recreated uh uh and they're uhjust but what we said um could
already speak like like us.
(15:08):
Yes, and fake fake videos couldbe made uh by us uh just thanks
to this uh like the 10 minutesthat we have spoken.
Uh but then the video thatwould be generated would not be
signed, and uh in the future uhuh you will be able to sign even
official videos of you uh withyour your uh your IDs uh and uh
(15:32):
strong maths.
So uh it will be possible todifferentiate an officially uh
an official uh video or uh audioof you uh or one which is
created by someone else.
Speaker 1 (15:46):
Oh, that's great.
Um the last season of mypodcast that I generated
completely with AI.
So I I made the the script andthen I I cloned my own voice and
make myself do the entire umthe entire episode without even
speaking.
Okay.
So it was quite fun, but alsoscary at the same time.
Uh my husband couldn't tell thedifference between when it was
(16:09):
me and when it was the AI, butmy mom could.
Because I think moms they havea better understanding.
And this is something it'sfascinating, but I love the idea
that you can you have thisextra layer of protection.
Like if in in the case someoneclones your voice or your
likeness, you will be able toidentify it as something that is
(16:30):
not authorized by you.
Yes, and you can still do it ifyou want to.
Speaker 2 (16:41):
You you know you sign on digitally.
Uh but this is uh and this isuh uh an outcome of uh uh some
European projects because Iparticipated to to many European
projects as part of Universityof Geneva leading uh this kind
of projects, and some you knowEurope has not been very good at
uh creating unicorns uh so farfor different different aspects,
(17:03):
but uh at least uh for allthose projects, like hundreds of
millions uh uh uh that workedon this uh uh ID and signature
uh uh technology, now it's beingembedded in many countries,
even at the notary, uh anddisqualified electronic
signatures actually are anoutcome of some of these early
projects I made in uh 15 yearsago, or even like 20 years ago.
(17:26):
Uh so from an identity point ofview, this is good because it's
also uh embedded into laws aswell.
And even Switzerland isfollowing because um you know
it's it's the relationship withthe yeah.
Speaker 1 (17:38):
Oh no, that's amazing.
I really like that idea, and Ithink it's it's uh it's the a
good answer of every technologythat we're seeing uh popping up
and how easy it is to clone uhpeople and and how easy also to
be deceived by that.
Do you you think it's the realperson behind it, but it's
actually AI?
Speaker 5 (17:55):
Yeah.
Speaker 1 (17:56):
So um you work uh with artists, cities, and
technologists.
When designing system fortrust, how do you balance
(18:34):
technical verification withhuman intuition?
Speaker 2 (18:39):
So uh human intuition is is a bit dangerous from a
trust point of view because umall my work is interdisciplinary
work.
So my PhD was of course havinguh uh computational models, but
based on the human notion oftrust from uh social science.
Um so I um you know read manyarticles on uh uh what was the
(19:05):
the notion of trust in uh indifferent countries, different
cultures.
Um and from that you know Ibuilt a computational trust
system.
Uh then of course you you knowyou not only have uh trust
research in social challenge butalso economics, psychology, and
so on.
Uh uh I also uh integrated somemedical aspects because we did
(19:29):
some brain computing uh researchon you know what happens in
your in your brain when yousomeone trusts someone else.
Oh wow.
And if quite if a system knowsuh how to trick what uh what is
what tends to make a humantrusting someone else.
So, for example, from a facepoint of view, if you have a
(19:52):
very symmetrical face, uh youwill have uh the human in front
of you will uh have a tendencyto trust you more than someone
else who has a not a symmetricalface.
Uh so for example, even AI, forexample, AI could create faces
which are sim very uh perfectlysymmetrical.
Uh and so um some science isdangerous intuition because uh
(20:18):
the uh you know uh thousand ofyears, uh hundreds of thousands
of years of evolution uh that wehave had in uh trying to uh uh
As human uh trust someone elseuh could be used against you.
Uh you know, so intuition is uhis not really like uh what I
(20:39):
would recommend for uhtrustworthy uh systems.
Uh so then of course technicaltrust is much more important.
As we said, like for example,um I you know your intuition
would think that you know avideo or someone that you see uh
seems trustworthy, uh but maybethere would be no, as we said,
(21:03):
signatures, qualified electronicsignatures from uh as part of
this video from a real human.
And so then you would have analarm saying that be careful, uh
it's very trustworthy becauseactually your your smart glasses
could watch uh also uh wouldhave watched the the video uh or
(21:24):
the person in front of you uhand say, okay, uh yes, it seems
uh very trustworthy, but thereis no uh digital signatures uh
of a human uh behind it.
So be very uh be very careful.
Speaker 1 (21:36):
Well that's scary.
So um by the fact that uhthere's a science behind behind
how we react to certain to thestimulus, so how we react to
people talking, how their facelooks, and you can map that in
the brain, then you can createalso that with uh uh which you
could create an AI that emulatesthat.
That's that's very interesting.
(21:57):
So it's it's interesting.
Speaker 2 (22:00):
Interesting for records, yes.
Speaker 1 (22:01):
Yeah, interesting is scary.
So reality can really be uhmanipulated against us, yes,
against what the instinct thatwe normally would have.
Speaker 2 (22:10):
And from a social point of view, you know, usually
trust is not perfect because uhuh me, I really use trust uh as
a human notion of trust, not asa uh a security, which is
usually you know, in security,uh either you have the wrong the
right password or the wrongone, or if it is uh signed
correctly or wrongly.
But even signatures, maybe in10 years' time when you have a
(22:32):
quantum computing, the currentalgorithms uh uh could be
broken.
So a signature is never uhsure, yeah, it's sure for today,
uh, but maybe in 10 years' timeuh you know it will be broken,
so it will be you will needsomething else.
You really uh should uh becareful uh when you when you uh
(22:59):
deal with uh information infront of you and without
technical trust.
Speaker 1 (23:03):
Okay, okay, that's that's interesting and scary at
the same time.
So your work has uh touchedeverything from offline crypto
transactions to augmentedhumans.
What's the red threadconnecting all of it?
Speaker 2 (23:19):
Also, I mean uh it's a bit opportunistic, but um
anyway, uh my my my research isabout trust.
Um and uh and so of course whenthere is a new system uh
emerging and and created, uh I'mgoing to look at if it is it
trustworthy, uh how could youhack it or uh how could you
(23:41):
improve uh its trustworthiness.
Um however, uh myself uh youknow the uh because you know we
are we are moving from the uhuse of a smartphone uh looking
at screens to uh as you said ofthe more a metaverse where you
you would have would you wouldaccess uh digital information
(24:04):
anytime anywhere uh by soundwith like uh an AI uh on your
smart glasses giving you uh somerecommendations or some
warnings uh uh by what you see,uh but in in augmented reality.
Uh and myself I'm uh also intouh what we call like this uh
(24:25):
biohacking from a personal pointof view.
Speaker 4 (24:27):
Okay.
Speaker 2 (24:28):
Uh so uh from augmented reality to uh
augmented your human body, youknow, um in 2010 we started
these conferences calledaugmented uh augmented human uh
uh international conferences,uh, where we were the focus was
more on the human body uh thanthe augmented reality
environment.
Speaker 1 (24:48):
Oh okay, to enhance uh our capabilities.
Speaker 2 (24:51):
Yes, or enhance your capabilities uh or uh retrieve
your capabilities because wehave uh some tracks which are
about uh rehabilitation, youknow.
Someone, for example, wouldlose uh a leg, uh, and then with
an exoskeleton and you know,maybe uh some uh uh different uh
techniques, uh sensors in thespine and so on, uh, could uh
(25:12):
gain regain uh a leg.
So it's not only augmented uhit's sometimes it also uh just
gaining back your abilities.
Speaker 1 (25:23):
Yeah, of course, especially with people with uh
with this kind of disabilitiesthat by accident or a disease
they lost uh uh an arm, a leg,or or or something else, and
they they can have uh uh anotherchance of of uh regaining
mobility, full mobility.
Speaker 2 (25:39):
And so, yes, almost in humans, then uh I look at uh
some trust issues.
Um but then it's very boldbecause we not only have um like
digital systems, but also uh weuh look at also uh medicine uh
supplements uh and so on.
Speaker 1 (25:59):
Anything that can improve uh the human uh so it's
really augmented from the broadsense, it's not only the the
little part of technology thatwe imagine.
Okay, okay.
No, it's it's it's a great workwhen you when you can really
change people's life to for thebest.
Speaker 5 (26:14):
Yes, that's right.
Speaker 1 (26:17):
So moving to um online reputation, um, which is
something that have shifted verymuch and it's it I think it's
it's uh it keeps changing.
So it was once just aboutreviews, so uh five stars, two
stars, and so forth over peoplecommentaries in in certain uh
(26:38):
blogs or online.
Now online reputation can shapeyour credit, your identity, um
also your future employment manytimes.
Um, and even innovationfunding.
Where is it going next?
Where do you think the onlinereputation is gonna take us?
Speaker 2 (26:57):
Um yeah, I mean, so online reputation will be also
uh in the real world because, asI said, uh you are going to
interact to have access to thedigital information anytime
anywhere.
So uh when you're going uhsomeone is going to be in front
of you uh with your smartglasses, maybe uh your uh AI
will uh tell you a number ofthings about this person.
(27:20):
So uh any uh yeah, the problemis the reputation will be not
only accessed online but also inaugmented terms, like anytime.
Speaker 1 (27:31):
Uh exactly I will see you when I will see when I will
see you.
Speaker 2 (27:35):
Yes, I will see uh uh all the good things you did
because you you convicted it canbe uh on the good side as well.
Yeah, of course.
You can do uh uh when someonegoogles about you, it could be
uh negative, but it could bealso positive and find good
information if you made the theeffort, like with this blog, for
example.
Thank you.
(27:55):
Uh qua this vlog, or like Idon't quite or interviews as you
as you call it.
Um then uh maybe for your nextpromotion uh it will help you a
lot to get because you will benoticed, you know.
Uh so in a good way.
In a good way, yes.
So online reputation is a youknow uh sometimes you you see
the we see the bad sides, but ofcourse the the one which are
(28:19):
most popular, uh then they gainfrom it.
Uh and yeah, it will be in realtime anytime uh anywhere.
So it will not be online, itwill be augmented the
reputation.
Okay, yeah, augmented becauseyou you will retrieve anything
that you uh which is public oreven which was not public but
stolen and put on the dark web.
Speaker 1 (28:39):
Oh, okay.
Speaker 2 (28:40):
Yeah, so um yeah.
Speaker 1 (28:43):
So the the hidden files that someone got from you.
Speaker 5 (28:46):
Yes.
Speaker 1 (28:47):
Okay, but that's it is it is good in a way of sense
of security as well.
For for example, I'm uh I thinkit is into a in a social
setting when you're meetingsomeone for the first time, or
when you want to make sure thatit's not a clustered killer or
something like that, it's goodto know their reputation.
Speaker 2 (29:06):
Yeah, it's good to know.
Uh but um yeah, I mean, youknow, you you you already have
some some tools like PMIS, uh,which uh with only one image of
you can retrieve all the publicimages uh of you online.
Okay, and I can tell you uh,well, I mean, um uh when you do
(29:27):
something, you you you should beprepared, especially online,
even if you use a digital tool.
Uh prepared that sometimesomeday it could be public.
Okay, okay.
Yeah, uh and our generation,you know, we are not we didn't
we didn't know about the risksthat we were taking by by by uh
(29:49):
you know all these generations.
Uh they don't were on TikTok,for example, when they publish
something, yeah, this is thereforever, it's been collected
already, yeah, and could beretrieved back.
Uh uh, and even uh yes, I meanuh even if you changed uh and
you have some TikTok videoswhere you were uh maybe drunk uh
yes, uh this will be uhretrieved uh and be seen by uh
(30:19):
in 10 years time.
Uh you will have uh so so youshould be prepared that uh I
anyway, even if you don't wantto do it digitally, you have so
many sensors everywhere uh thatuh we should be prepared to be
to accept that uh we are beingwatched, we are being watched
(30:41):
the full time.
And so uh well I mean you it'sdifficult sometimes uh to to to
restrain yourself or to yeah ofcourse but um uh yeah some but
then then uh yeah yeah and thenit will be very difficult to to
to protect it.
Speaker 1 (30:56):
So yeah, it's um it's it's interesting in that way
that uh there's a lot of um umbelief that privacy itself is
not gonna be something that isuh the same sense that we have
it right now is not gonna existanymore.
Because we will be constantlybeing observed, either either
monitored by devices that arearound us or by the the the ones
(31:19):
that are outside in the publicwhen we go outside.
So there's never gonna be apure privacy moment um in the
near future.
Speaker 2 (31:27):
Yes, it's it will be possible uh to remain private,
uh, but uh it will be uhdifficult uh and this is
difficult to remain private fora long time.
Speaker 1 (31:39):
Yeah, yeah.
Well, interesting.
So you explored uh the legalweight of digital signatures on
NFTs.
Do you believe we're movingtowards enforceable trust in the
metaverse?
Speaker 2 (31:53):
Although uh trust as a human notion of trust uh is
not enforceable.
Um this maybe you could have uhuh a number between zero and
one hundred.
Um but you see you cannot forcethe whole world to trust you.
So you see, from a world pointof view, you cannot force
(32:15):
everybody uh because the worldis uh fortunately is still uh
has still some independence.
Yeah, uh so uh from a socialbecause trust uh is also linked
to your social network.
Uh from the sociology point ofview, uh we always talk about
those social networks, theyexisted before, uh of course,
when the when you know uh anumber of uh you have your
(32:39):
parents, your family, yourfriends, and then they know
other people, uh you you havethis kind of uh social networks,
they existed for a long time.
Uh now with the social networksyou have maybe weaker uh links
because uh, for example, onInstagram, so the new generation
we are doing some research atthe moment at University of
Geneva.
Uh before it was more like thethe brain is more made to really
(33:01):
have close quite some a circleof uh 150 people.
But uh it's not uncommon thatuh the younger generations from
15 to 30 years old now, uh theyhave uh 500 uh for the men to
1000 uh weak followers or orcontacts, you know, on those
(33:24):
Instagrams or social networks.
Uh so it's getting uh it's it'sgetting broader.
And anyway, uh by you know thetrust also involves
recommendations from others.
Um so when you build a trustmechanism when reviews, for
example, okay, you could havereviews which are non-linked to
(33:46):
you, but if they are if youryour close friends uh also
provide uh reviews, uh andsometimes at some some stage
they they try to add this intouh uh into triple advisor.
For example, you could see uhyou could connect with your
Facebook account, you know youhad a Facebook Connect, but you
know now everybody uh is afraidof Facebook Connect.
(34:06):
So but at some stage it worked,and then you were seeing uh
reviews of hotels of yourfriends.
Speaker 4 (34:11):
Oh, okay.
Speaker 2 (34:12):
And so uh somehow, of course, uh you tend to trust
more the people, the closefriends than than uh than others
if you uh if you if if you knowthat they are uh stable.
Speaker 1 (34:28):
Yeah, because it it it makes more reference to the
same person, so moreconfirmation about who the
person is and how it be theybehave.
Speaker 2 (34:36):
So you see, you cannot enforce.
Uh you can try to influencebecause you can try to build on
your social networks.
You can as a hacker, you cantry to uh uh become the friend
because if someone is very uhhard into uh getting into uh you
could try to select to toanalyze the social network of
the uh of that person and try tosee uh well position uh
(34:59):
profiles which seems to be easyto to trick into trusting you.
And because then after youbecome a friend of this person,
uh and someone, for example, onLinkedIn, you know, is trying to
uh a hacker is trying to getconnected into your so linked uh
social networks, professionalnetworks, but it's a social
(35:21):
networks, uh you would try to uhbe already a contact of someone
you know, and you can see youknow, oh this guy knows this
person.
Uh and uh by that, if there isno double check, uh you would
see a request from this uh otherperson or contact of uh a
professional contact you know,so you it will facilitate this
(35:47):
request uh more easily than thanum if uh it was there was no
connection.
Speaker 1 (35:53):
You have no one in common.
So how how did he get to me ifI have no one in common?
Speaker 2 (35:57):
Yes.
So uh and then you cannot conyou cannot have a social
contract network of the wholeworld.
That's the same.
Uh uh the farthest you go, uhthe less influence they have.
Um but then you you you can tryto influence, uh, that's what I
say in my courses.
Uh, online reputation at theMedia Lab in uh University of
Geneva.
We have a uh online reputationmonitoring um course, and uh so
(36:22):
it's not enforceable, it's um uhyou can influence it, but not
enforce it.
Speaker 1 (36:26):
Okay, interesting.
I would I would like to knowwhat it comes out if you put my
name.
A lot of things about IPprobably.
So with the rise of generativeAI, which is in everyone's mind
right now, um, what's your takeon trust boundaries?
How much should we reveal aboutwhat's AI made?
(36:49):
So if I generate uh somethingwith AI, should in every every
moment I should always declarethat I made it with AI, or can I
just make some things andwithout saying that it was AI
generated?
And how would that affect thetrust?
So the people will expect thatfull transparency.
I would like to know everythingthat is AI generated, I would
(37:11):
like to know beforehand, or itcould it could be more flexible,
like maybe not I don't need toknow everything, just things
aren't sensitive, or that canaffect me.
Speaker 2 (37:21):
The thing is uh AI is going to be everywhere anytime,
anytime any anywhere.
Uh you know, as I said withaudio or even like uh in in in
other terms.
So um you are it can generate,but it can also recommend you
things, so uh it will inf itwill influence uh anything you
(37:43):
decide.
Um so to me, I mean it's morelike um we are more um we are
more interested in in knowinghow much human work has been put
into a product or uh art or uhactivity uh than knowing that AI
(38:03):
was used because uh in 10years' time uh AI will have been
used for everything almost.
Speaker 1 (38:11):
Okay, it's gonna be the norm.
Speaker 2 (38:13):
So you would have a warning for uh for everything
constant everything.
So uh but then uh on the otherhand, uh how much human work has
been uh done for a specific uhproduct or task uh is uh more
interesting to me.
We are working on uh as part inGeneva as part of a project
called House of Good AI.
(38:34):
Uh, we have also a good AI uhuh society uh when we do some
meetups, and we discussed um uhthat it would be interesting not
having a uh human-made labelbecause uh if you craft you
craft uh uh like for wood, uh ofcourse it could be be 100%
(38:55):
human-made.
But if it's a document or ifit's anything else, um uh you
you you would use already uh acomputer which is not human-made
or uh or some some someprograms which are not
human-made because uh some AIhas already uh participated to
it.
So um it would be uh moreinteresting to uh than to have
(39:21):
uh 100% human-made, to have uhuh and what what we call uh uh
in house of good AI we we havethis way of uh uh allowing any
any people uh to uh have theirown cryptocurrency, which one uh
cryptocurrency of Jean-Marc isequivalent to one hour of my
(39:42):
time, and you would have uh uhyour own token uh uh which is uh
one hour of your time that youhave lived, and uh you could
decide uh how much of thosetokens are are uh being spent
for some activities.
So, for example, in art, uh apainter could say, Okay, uh this
(40:03):
uh painting I made four hours,uh it took four hours of my
time, but also an AI generatedvideo, even if you uh used uh AI
generation, it it took some ofit of your time to uh to to
spend.
So you could say, Okay, it tookme 10 minutes to uh to write
the prompt and so on.
Uh and then uh instead ofhaving a percentage, because
(40:24):
it's a percentage is difficultto really um uh compute or to
have the right information uh bycounting the number of tokens
that the product has uh has beenspent for for this um then you
would have an estimation uhestimation of uh uh how many
hours of work of human work hasbeen spent uh to this product or
(40:47):
even like just paid asemployment uh sometimes we we we
we are laughing because um uhyou know some people they say
that they have a bullshit uh jobuh meaning that you know they
they are not doing much for thecompany but uh I mean they are
here and they are paid uh so uhinstead of having a uh a minimum
(41:08):
uh salary that uh uh tocompensate that you don't have a
job and um and you are onlypaid uh an amount maybe an AI
could pay you uh just to spendsome of your time tokens for uh
uh for having a a per like anumber of uh human work paid uh
(41:29):
even if you did nothing but atleast uh uh maybe close to your
your your social network to givesome uh value uh other value uh
human value to to the productthat or service that the the uh
they buy so you're thinkingabout that this uh very long
debate about the the generalminimum wage that people are
(41:52):
talking about that everyoneshould receive a minimum wage if
even if they're not uhespecially those that are not um
able to work just yet or theyoung people that uh still don't
are not enter the workforce uminstead of having that just
because of existing then theidea would be if you are putting
yourself into um work as ahuman hour for for AI purposes
(42:17):
would be a good way to earn thisyeah an AI could could choose
you know uh like some people uhit's kind of hiring but maybe
doing nothing it could be moreflexible yeah could be do could
be used to uh maybe uh marketthe the product to some to their
social uh circles okay uhbecause maybe the AI has spotted
that uh uh this is uh this is awealthy uh family uh um and so
(42:46):
uh yeah in this way I mean atleast uh it's a bit clearer I
mean also well positioned peoplecould get more uh paid more
could be paid more for that yeahinteresting so so it's uh
there's a lot of uh there's alot of ideas here to um to
address that this um the oneabout most things will be AI
(43:08):
generated anyways in the nearfuture so the labeling should be
for human input not for AI yeahuh uh uh uh you would have uh
on the label like uh the numberof hours that have been uh spent
and proven you know that thisis secure on uh the the House of
Good AI have a we have a um uhan an attack resistant uh
(43:29):
algorithm uh also put on apublic blockchain yeah uh and I
mean cheating is uh uh we'remaking mechanisms uh against
cheating so it would really uhkind of uh it would be public
information uh that any uh anyAI could uh retrieve on on uh on
(43:50):
the blockchain.
Okay interesting and moving onto the next question how do we
prevent algorithm trust frombecoming automated bias I know
(44:16):
um as I said my vision of trustis more like a human notion of
trust and uh uh it will tend tobe biased of uh based on your
social network and and historyand so on uh and trust is not
global as I said uh twodifferent persons could have a
(44:38):
very different trust value insome something else due to their
social network and and pastexperience and recommendations
uh and anything that that thathappened so um um yeah um so
there's there's always gonna beuh bias and this it's yeah I
(44:59):
mean to me it's by by design itis biased okay because we are by
design yes but on the other onethe uh the bias uh will uh uh
tend to uh to be uh to makedecisions or um to to be qu
(45:20):
would would be uh used uh andand and should the the goal of
of this system or this trust uhmechanism is to have the the the
most the optimal risk rewardyou know so um usually I mean
the bias is the ideally uh ifyou have good trust mechanism uh
(45:41):
metric mechanism uh would um bein your favor because it could
be biased against you um so itwould yeah yeah so we it it's
something that is uh is humannature to to have uh different
judgment and to have differentapproaches yeah uh if if you
take trust as a human notion oftrust uh it is biased uh you
(46:05):
would you you will tend toprotect more your family than uh
than strangers in many culturesalmost all cultures because uh
you know the maybe the ones thatwere not protecting their
families they they disappearedalready due to evolution makes
sense if you could embed onetrust metric into everyday
(46:26):
platforms what would it measurewhat what it would measure um to
me I mean it it would be uhmaybe based on one of my early
(46:58):
work uh that I uh designed uh itwas called the trusto um and
really is the the would be thehuman trust you know the if you
have a human component commoncomponent in your system would
be uh the human uh uh trust uhof the current users you know
(47:22):
the the trust of the of thishuman uh regarding the mechanism
or system uh which is being uhevaluated uh so it would be the
human notion of trust.
Okay okay then what triggers usto trust uh I mean as as the
(47:43):
component in in the distrustmetric okay uh it would be
important to take into accountperfect so now we have the the
fun part of the episode um Iwill ask you um one um I I'll
(48:07):
ask you to pick one of these twostatements so first one is
verify synonyms or totaltransparency or verify you mean
like uh uh that it there there'sstill a synonym you don't know
exactly who is behind but it'sverified that it's a person but
(48:29):
you don't know the name of theperson then the second one is
total transparency of course uhcome to me at least verified so
it's gonna be uh um flower oneto three if you know it's a
verified user but you don't knowthe name of the person okay
perfect NFT has proof ofownership or traditional IP
(48:51):
registration uh NFT has proof ofownership maybe sign NFTs you
know one global language poweredby translationships or keep
every language alive I mean keepevery language alive but uh as
we said uh AI will translateeverything uh very easily so uh
(49:14):
so a combination yeah I mean nobut uh it's it's very
interesting I think to have uhthe diverse uh diversity into in
into in the languages um butyou know AI will translate uh
you know automatically what wehear so um but then maybe it
(49:35):
even like somehow it might itmight protect maybe more like uh
uh less spoken languagesbecause uh usually what happens
is that you need to learn thenational language to to be able
to work in in the nation um butthen if you know uh at that
(49:59):
stage if when you speak yournative language or like local
language uh dialect uh it'sautomatically understood by the
person in front then uh you haveless uh you you need less to uh
to to forget about your dialectand and can keep it it will
keep them alive because then youwill be able to use it really
(50:19):
without uh um without withouthaving to go to and learn the
other one yes so I mean maybethat that's the other way I mean
maybe it's about it's positive.
I love that I love that so touse the AI has a tool to
preserve the language yeah sodecentralized data governance or
citizen led data trust sopeople will run the data or a
(50:42):
decentralized way to run thedata the the citizen would um
would own their own data or itwould be centralized yes uh no
it should be uh you should quiteideally your privacy and your
private your your data shouldyou should uh uh be able to own
(51:05):
it and decide when you uh wouldhave to share it uh yes I'm more
in in this favor like uh forfor this than than uh of course
having a centralized uhmonitoring system yes so citizen
led immutable records ofeverything or selective
(51:27):
forgetfulness by design do youhave the right to be forgotten
or not um but it depends I meanum you mean after you you you
you uh you die yes but um umdifficult to to to uh to uh to
(51:52):
answer us those questions yeahthat's that's the idea because
they both of them have merit yesum I think for like from a
freedom point of view uh onceyou are died you you you should
have the the right to be uh tobe forgotten because the the
(52:14):
thing is that if you didmalicious activities uh and you
are still alive and they areforgotten uh well I mean uh you
cannot be uh judged uh for forwhat you uh what you did but
once you are you are dead so Ithink uh yes I mean uh
forgetting uh choosing to thatyour data is erased after you
(52:35):
you die uh I think for me wouldbe fine before uh especially for
criminal activities yeah ofcourse those should be kept
because there's public interestfor them yeah and then they were
they are they were malicious uhyeah yeah okay so a combination
of both transparent uhtransparent governance or
(52:57):
invisible protocols that workwell it depends for what also uh
uh so some system you like uhtransparency for example maybe
for governments it would be niceto see exactly uh what you paid
for how they reached thatdecision how they how much they
invested in that uh did it go tocorruption or not how much went
(53:20):
to corruption okay uh but um uhyeah so it depends on on on
which system you are you're youare talking about for the public
uh goods uh yes I mean thegovernance should be transparent
uh when it matters of uhprivate uh uh aspects uh as I
said ideally privacy uh shouldremain but is it possible yeah
(53:43):
okay AI trust ratings issued bypeers or issue by regulators uh
could you repeat the first onesorry I mean uh so AI trust
ratings yes either issue bypeers or issue by regulators do
you trust more what the peoplethink in general the general
(54:05):
public think or what a regulatoruh decides that if if it's
truthworthy or not uh but thenwhy why do we say AI trust?
Because uh so it's a let's sayit's an AI tool and they're
ranked the option is is eitherranked uh by the public or it's
ranked by the regulators by thegovernment who would you trust
(54:27):
more to rank the AL2 you thetool you need a mix of both
because um uh in many uh in manyaspects uh you have a
competition between theattackers and the defenders and
your government uh as much moneythan uh uh than yourself
usually uh so they can go intomore details uh about some
(54:51):
technical aspects um and again Ithink it we need both yeah so
good so every everyone will havetheir own role on on assessing
the tool yeah you you you youyou need to have your own
thinking uh of course uh butthen as I said uh sometimes um
you wouldn't have enough to uhto assess you know the what is
(55:13):
in front of you and so uh thenbecause it's your government uh
I mean if you then you also needhave to trust your government
of course but uh usually I meanuh in democracy it's uh it
should be uh in your favor uh toum then you need the more okay
uh bigger uh players to toassess what is in front of you
(55:33):
uh to spend a lot of resourcesand and knowledge uh and somehow
uh because it's your governmentuh you tend to give it a bit
more trust than uh if it was thecompeting government you know
okay makes sense ecosystemswhere trust can be restored or
(55:55):
ones where it is lost it isirreversible no you can you can
always uh uh try to recovertrust but this is uh difficult
you can lose trust you know ittakes a long time to build trust
in the human uh world um ittakes more time I mean to build
(56:15):
trust than to lose it yeah it'strue you can lose it in one
second you can upload one memoryto the internet private forever
or public for everyone um couldyou repeat again this is a bit
(56:36):
uh sci-fi kind of thing you canupload one memory to the
internet one you mean one memoryso one memory from your life to
the internet and you willallload it to be private or to
be public no but of course Imean the the same uh depends on
uh it should be good to thatit's it's kept uh as a as a as a
(56:58):
souvenir uh memory uh but thenyou should be able to to choose
if it's private or not I meanthen again as I said now as soon
as you put something online uheven on a supposingly secure uh
service uh when it's digitalizeduh uh it can become public so
um but ideally uh you should yesthe for me I'm uh for the human
(57:24):
privacy and so uh should beable to choose uh uh if it's
private or not.
Speaker 1 (57:30):
Okay, so for everyone's to choose.
Speaker 2 (57:33):
Yes.
Speaker 1 (57:34):
Okay now uh take the palette okay and this is this is
part of the game of true orfuturist.
Okay.
So I will give you a statementand you choose that is true that
is meaning that it's happeningright now or futuristic that is
going to happen in a hundredyears or it's never going to
happen.
Okay ready?
(57:55):
Yes NFTs will be legallyequivalent to a notarial
signature so like a notary umyeah also then uh true true
augmented humans will carrydynamic trust layers visible
through augmented reality so wewhat we were talking about
(58:19):
futuristic is what like uh fiveyears ten years or more no no uh
50 years uh more more morefurther in the future uh true
true okay because we were justtalking about that before more
in ten years time okay ten yearsokay interesting web3 platforms
will have ethics protocol votedon on by users um yeah I think
(58:48):
true true interoperable identitywill become a requirement for
cross platform login no morepasswords and usernames um yes I
mean the interoperable uh uh II access this is true also you
(59:10):
know no problem yeah so before50 years of course before 50
years I I would love to be ableto forget all the usernames and
passwords that I have to createevery time but then you might
lose some privacy yeah that'strue but you can never uh
there's uh my husband says thathe doesn't believe in progress
because whenever you gainsomething you lose something
else so there's always gonna bea bargain you always got there's
(59:33):
always gonna be a bargain soit's interesting because it's
true it's that you you have toyou don't have to hurl out of
anymore of creating usernamesand passwords but then you have
the issue that someone can takeit and then you're you're locked
out of everything.
Speaker 2 (59:48):
Okay conflict resolution will shift from legal
courts to smart trust ecosystemI mean the lawyers are are very
uh strong even if AI is isgetting uh smarter um so I think
(01:00:08):
uh I may think in future westill need uh uh lawyers okay
that's so so futuristic it'sgood so we still we still need
it okay your rowsing behaviorwill be used to compute
trustworthiness by insurancecompanies
Speaker 1 (01:00:31):
Well this is this is uh ah well usually the is it
really important for them to toto know uh your trust or sinus
is more like uh your elf no orif if they from that they can
get real information that ifyou're googling uh something
that is uh I mean or you'rebehaving recklessly.
Speaker 2 (01:00:54):
I mean uh if they if they succeed uh it's the it's a
bit difficult.
Um but um yeah more and it'seasier and easier uh uh as I
said because we have more andmore information on online uh or
digital.
Speaker 1 (01:01:10):
Okay.
AI will have to explain itselfbefore making a decision about
you.
Speaker 2 (01:01:24):
In 10 years' time, I mean of course uh it will be
already here.
I mean they will uh help us toto make decisions, so I mean
it's true.
Um but sometimes I mean you youyou know uh people they when
they they they have to to makesecurity decisions for example
uh as we have seen for cookieson on website and so on, usually
(01:01:46):
I mean uh they don't have timeto to read the the the licenses,
uh they don't have theknowledge, for example, uh uh or
they don't have the money uh topay for some uh strong AIs or
mechanisms to help them.
Um and so because they don'thave time, uh for some decisions
(01:02:07):
the AI, I guess uh people willuh clearly let uh AI deciding
deciding for them because uhthey don't have time and they
are sometimes maybe also uh lazysometimes and then of course
there are risks and then maybeuh but then explaining
themselves they they will beable to um to do that in 10
(01:02:28):
years' time.
Speaker 1 (01:02:31):
We'll need a license to use powerful AI tools like we
do with cars.
With cars just or with amedical license, for example,
engineering uh um certification,so uh AI tools that can really
create something that could posea harm to the public.
(01:02:53):
Would that require also like acertification, a license, or an
official authorization to usethem?
Speaker 2 (01:03:01):
Well, I think it's it's true because that that's
what they are trying to do now,but uh if AI continues to to
make progress uh like that, uhin 10 years' time or 20 years'
time, we might have uh superintelligence.
Uh uh and so uh this will be abit uh you know uh we won't
(01:03:23):
understand you know what what'sgoing on, and and so uh it's
gonna be beyond us.
I mean um so um now it's acase, but then it might
disappear.
Speaker 1 (01:03:35):
Okay.
You receive personalizedpersonalized laws based on your
online behavior.
Personalized laws that applyonly to you.
Speaker 2 (01:03:50):
Well, yeah, you know, I'm not uh I'm not uh uh into
legal uh like nanny state kindof thing.
Uh no, I uh no uh uh well Iguess law uh should be a bit uh
like for a country uh size orlike uh it's uh uh a kind of a
uh regulation uh uh uh for acommunity of people.
(01:04:11):
So only for you.
I don't see it.
So I mean uh because yeah, youknow it's not even now, not in
the future, it's never no, neverso yeah, it's never like this.
Speaker 1 (01:04:24):
Final question.
You'll take person you willtake personality test before
being allowed to use certainplatforms to see if you're
messed at a stability or youhave the connectivity uh or you
have the development.
Speaker 2 (01:04:40):
I think this is already known.
I mean, uh uh, but in thefuture, as I said, we we won't
use platforms.
Uh we will interact with themetaverse like uh uh
automatically, like everydaylife.
Yeah, anytime anywhere.
It's kind of uh what we callVaser, for example, uh was uh
calling it uh ubiquitouscomputing.
Uh and we already worked on onthis, like a global global
(01:05:03):
computing, you be ubiquitouscomputing.
He was talking of thedisappearing computer.
Um yeah, and then everythingwill we we we'll we won't have
platforms anymore.
Speaker 1 (01:05:14):
Um so it's gonna be the everyday life?
Speaker 2 (01:05:16):
Yes.
Speaker 1 (01:05:17):
Okay.
Is this so true?
Speaker 5 (01:05:21):
Uh yes.
Speaker 1 (01:05:26):
Okay, so now um the question to wrap the episode.
Has technology gets faster andmore visible?
Was one human truth you believeevery trust system should
protect?
Speaker 3 (01:05:47):
Human truth.
Speaker 2 (01:05:49):
Um, this is very narrow one, but uh let's say uh
I would say uh freedom.
Speaker 1 (01:06:01):
Yeah, I love that.
It's it's the it's the core ofhumanity.
Thank you so much, and Marcushas been an amazing
conversation.
Speaker 2 (01:06:12):
Sometimes difficult, but uh but very well thought,
you know, like you you spent alot of time to to to think about
these uh these concepts andeven yourself.
I mean you you uh you have avery good you did a very good
brainstorming, so it wasenjoyable.
Speaker 1 (01:06:25):
Yeah, that's very sweet, thank you so much.
No, I had a great time.
It is it was very, veryinteresting, scary in a good
way.
Because the future is is uh issomething that is um enticing,
and we always want to know whatis going to happen next, but it
can be also scary if uh if thepeople the right people are not
(01:06:47):
making the right decisions.
So yeah.
Yeah.
So thank you so much forreminding us uh that trust is a
framework, and in a digitalworld, that framework must be
designed, debated, and defended.
Your work shows that what wetrust shapes what we build.
And if we want technologiesthat empower, not just impress,
(01:07:09):
it starts with a reputation thatis earned, not bought.
Because the future isn't justdecentralized, it is deeply
personal.
Speaker 2 (01:07:18):
Thank you.
Speaker 1 (01:07:19):
Thank you.
Speaker (01:07:24):
Thank you for listening to Intangibilia, the podcast of
Intangible Law.
Plain talk about intellectualproperty.
Did you like what we talkedtoday?
Please share with your network.
Do you want to learn more aboutintellectual property?
Subscribe now on your favoritepodcast player.
Follow us on Instagram,Facebook, LinkedIn, and Twitter.
(01:07:44):
Visit our websitewww.intangibia.com.
Copyright Leticia Caminero2020.
All rights reserved.
This podcast is provided forinformation purposes only.
My my my research is about trust.
And so of course when there isa new system uh emerging and
created, I'm going to look at ifis it trustworthy?
How could you hack it or uh howcould you improve uh its
trustworthiness?
Speaker 1 (00:18):
Um welcome to our podcast.
Speaker 2 (00:29):
Thank you for the introduction.
Speaker 1 (00:31):
Um so the first question is please uh let us
know who you are and how um youcame to be this uh IP expert.
Speaker 2 (00:40):
I mean um uh I've been working on uh decentralized
trust since uh 2000.
I started my research atTrinity College Dublin, I did my
PhD there on this topic.
Uh then after I joined theUniversity of Geneva, where I uh
teach uh online reputation aspart of the Media Lab of uh the
Faculty of Um Social UhSciences, and um uh as well as
(01:07):
directing the certificate ofadvanced studies on blockchain
technology, which is also uh atechnical trust mechanism.
Speaker 1 (01:16):
Yeah, perfect.
Um so you have been working ondigital trust since the early
2000s, that's uh quite a longtime.
Um, what has changed the mostin how trust is built online?
Speaker 2 (01:31):
Uh uh so online uh for the last 20 years uh it's
got it's gotten worse uh becauseuh the technology uh hasn't
changed much from a securitypoint of view.
Um and uh the hackers uh haveno uh know better how to attack
(01:54):
those systems where security hasnot been enough uh uh sought
and and uh invested in.
Uh so uh now there are moreattacks and and security is less
uh as before, uh almost asbefore.
Um but in the last years we wehave seen coming up uh stronger
(02:17):
mechanisms, uh maybe linked touh chips on identity cards where
you could really prove your uhreal identity uh as a human uh
citizen of a country uh online,and uh it's going to help
securing more uh the internet.
Uh but so far no, I mean forthe last uh 20 years uh not much
(02:40):
has been done because uh notsecurity has not been taken into
uh consideration as a strategicuh uh asset.
Um even countries uh have noteven invested in in uh digital
uh technologies.
Uh some countries have, like uhthe USA or or China, but uh
(03:00):
Europe, Europe, for example, oreven Switzerland, they are
talking about uh digitalsovereignty, but uh they have
invested nothing in the coretechnologies, uh and so uh it's
very weak now.
Speaker 1 (03:13):
Okay, so the focus has been to digitalize and to
become more modern, but notthinking about the risk of that
uh digitalization of being moreexposed to the risk online.
Speaker 2 (03:25):
Yes, and and we see different risks because they are
like independent hackers butalso uh like aggressive
countries now.
Oh okay.
Um well, and so uh the internetis uh is not uh uh always uh a
playground where we only uhplay.
Speaker 1 (03:43):
Everyone is there, the good and the bad.
Okay, makes sense.
Makes sense.
So you have helped create oneof the first academic programs
on blockchain and decentralizedapps in Geneva.
Why is education a criticalpillar in shaping through tech?
Speaker 2 (04:01):
This is very important because otherwise you
you seem to uh to know how touse the those systems because
they are very no uh they areuser-friendly, uh, but you don't
know the underlying uh risks.
Uh and so um it's important forthat in each country uh a good
number of people uh know uh thedetails uh and know how to
(04:24):
protect uh their citizens.
Speaker 1 (04:26):
So education is key in this sense.
Speaker 2 (04:28):
Yes, if you don't know the details.
I mean uh uh you rely on onother countries who would
provide you uh we will uh whichwill provide you some um like
technologies that they say aresecure, uh but you don't know.
Uh you you need to to do peerreview and also internal peer
review.
Speaker 1 (04:47):
To verify yourself just in case.
Speaker 2 (04:49):
Yes.
Speaker 1 (04:50):
Yeah, makes sense.
So and also to make sure thatyou learned what the technology
behind it and you understand umwhat is what exactly it's about
in order to be able to regulateor to use it properly.
Speaker 2 (05:02):
Yes, especially when you have hackers or aggressive
countries, you know, who providesome technologies.
Speaker 1 (05:07):
Yeah, makes sense.
So from CV CBDCs uh to NFTs,what's the one sample um where
decentralization improved orchallenged public trust?
Speaker 2 (05:22):
This is very broad because CBDC are central bank
digital currencies, uh, whichmeans that uh the central
central bank would uh create uh,for example, in Switzerland a
digital franc or in the US adigital dollar.
Uh it's another way of paying.
Uh now we have coins, uh, wehave uh banknotes.
Um in some countries we havecheck.
(05:44):
Uh then you can pay also withapplication, uh like uh online
payment.
Uh you have cryptocurrencies,and here uh uh the idea would be
that you have uh uh a money uhwhich is uh uh also provided as
another uh with another means asa digital currency, but it
(06:04):
would be like a real dollar oruh real uh Swiss franc.
There are many different waysto do it.
Uh usually uh we're talkingabout cryptocurrencies, like uh
an equivalent of uh it would notBitcoin but uh a dollar that
would be created uh as acryptocurrency by the Fed or uh
(06:28):
in Switzerland the SwissNational Bank.
Uh but it's not the only way todo it because on the blockchain
cryptocurrencies are also on ablockchain and they are tracked,
and from a privacy point ofview, it's not so so good.
So there are other uh ways todo it.
So uh CBDC are centralizedbecause they are uh provided by
(06:49):
uh a central bank of a country.
So it's very different than uhNFTs, which are non-fungible
tokens, uh, which are usuallyused to uh yes uh you you you in
your crypto hardware wallet orcrypto software wallet, you can
have uh cryptocurrencies butalso those uh NFTs, non-fungible
(07:10):
tokens, uh, that can be linkedto uh a certificate of ownership
of uh piece of art or um ofsomething in the real world,
which is um supposed to be umlike uh real ownership.
Okay.
Uh one piece which is missingin current NFTs is that okay, on
(07:33):
the blockchain this is secure,this is uh really in your uh
when it's it it's written in theblockchain uh and you own this
uh NFT in your crypto wallet.
Um yes, you have it by and it'sprotected by maths, by by
cryptography, uh on theblockchain, in your uh crypto
wallet, but lots of judges uh orlaws uh in different countries
(07:57):
they don't know aboutblockchain, so it's very
difficult to explain to them uhyou know uh yes, this is my
ownership because this is my uhcrypto wallet.
Uh so uh it's difficult toprove really that uh you own it
really in the real world.
Um so then we have worked withuh International
Telecommunication Union uh for anew standard called the Sign
(08:18):
NFTs, meaning that with the NFTuh it's inside your crypto
wallet, but also uh you have auh digital signature which is uh
uh of very high uh strength,qualified electronic signature.
And uh then when you open thisuh NFT, you have also a link,
(08:39):
for example, for a PDF, and thisPDF will be signed digitally
by, for example, the artist, uhand it's equivalent legally as
uh uh the handwritten uhsignature um by law in
Switzerland or Europe, and Ithink there are some uh it will
be also recognized in other uhjurisdictions.
(09:02):
So um, yes, I mean so then yousee from CBDC, which is
centralized, to NFTs that can uha piece of art can be created
by any artist in the world.
Uh you have a big differencebecause it's uh centralized and
NFTs are more in a decentralizedway because anybody can create
NFT, uh a central bank currencycan only be created by a central
(09:24):
bank, right?
Um, and so uh the um last pointof your question was like so
it's um what is the thechallenge or the improvement on
public trust?
Speaker 1 (09:36):
From what you're saying, the one with the the the
extra proof for the NFT, that'sgreat for trust because then
you have an authority that isconfirming the validity of of
this um of the of the art or thework behind the NFT.
Um and what the other one aboutthe um the the one that is uh
(09:56):
managed by the banks, you havethe banks, people usually trust
banks in the majority.
Um what challenges do you seein this kind of because uh in
this kind of development?
Because um if the um coin isgonna get all digitalized, um it
can get uh as well, it can getuh lost or it can get uh uh
(10:21):
break down or someone can attackthat you can lose everything
because it's only digital.
Um so in this case, how do youbuild trust on that?
Speaker 2 (10:30):
Yes, alors so um uh from uh what I understand is
that also from the public pointof view, um then all of this is
is very new.
Uh and so uh there's a bigquestion in uh the population or
the citizen trusting uh thesenew cryptocurrencies.
Uh so from uh a CBDC point ofview, there is uh for the normal
(10:56):
citizens who uh who try totrust in this system, if it's
provided by the central bank, uhthen already from a legal point
of view there is no uh trustissue because this is a legal
tender uh um they are regulatedand so you know this is a uh um
a governmental organization.
(11:18):
Um and so then uh uh from thispoint of view, uh there is less
the the trust question is lessimportant because this is uh uh
provided by uh a by law uh aninstitution that uh says this is
uh uh if you have such CBDC inyour crypto wallet, then you you
you own this uh this money andthis is an official money.
(11:41):
Um so here there is a less lessrisk.
Uh from NFTs, then you you youalso have this education for for
uh uh normal people you know inSwitzerland or in developed
countries, you have between 10to 15 percent of the population
who have already bought somecryptocurrencies, uh, but
(12:02):
sometimes via intermediaries.
Um, you can you you don't needyour own crypto wallet, you can
buy it from some banks.
Uh for example.
Uh in Geneva, you you have aArab Bank Switzerland, for
example, uh you have Swisscoat.
Um and uh so even some of thesepeople don't understand really
(12:24):
uh don't have their own cryptowallet, so uh you have a lot of
of uh uh education to do to forthem to really be uh uh used to
use this new technology.
As the web, you know, it tooklike 20 years, for example,
before there was a large scaleadoption.
(12:45):
Uh and then also for the cryptowallets you need uh you need uh
other components for at theuser interface level that will,
as you said, help not to losethe the the keys for the uh the
crypto wallets, uh to recoverit.
Uh and um for the next fiveyears uh there are some
(13:10):
different working groups workingon uh bringing um easier user
interface for cryptocurrenciesand NFTs.
And uh already, as I said, NFTsuh they miss uh already uh an
official digital signature ofthe artist, or you know, like
the the company that would uhsell um um let's say a flat uh
(13:36):
or uh house uh ownershipcertificate.
Uh you need to add somequalified electronic signatures.
But the good thing is that uhnow in Europe and Ireland you
have these.
This is not part of blockchain,but is this is another part
technology technological part.
Uh most you uh most uh nationalIDs will have chips uh uh uh
(14:01):
that will be uh certified by anumber of trusted parties.
Uh in in Switzerland, forexample, for the arts, uh
Swisscom is uh is one of thesetrusted third parties.
Um as I said at the beginning,now we we are moving into uh uh
maybe an era where it's a biteasier to be sure who uh uh who
(14:26):
is talking in front of you, evenif you have all these AIs that
you could impersonate uh theface, the image, and so on, they
will not be able to impersonatethe mats and the signatures.
Uh so um uh yeah, at themoment, even with AI, it's even
uh more difficult because ifhackers are uh building AI of
(14:47):
people uh with a voice, theface, yeah, all the information
which are uh known about us.
Uh this interview, for example,uh already uh an AI could be uh
recreated uh uh and they're uhjust but what we said um could
already speak like like us.
(15:08):
Yes, and fake fake videos couldbe made uh by us uh just thanks
to this uh like the 10 minutesthat we have spoken.
Uh but then the video thatwould be generated would not be
signed, and uh in the future uhuh you will be able to sign even
official videos of you uh withyour your uh your IDs uh and uh
(15:32):
strong maths.
So uh it will be possible todifferentiate an officially uh
an official uh video or uh audioof you uh or one which is
created by someone else.
Speaker 1 (15:46):
Oh, that's great.
Um the last season of mypodcast that I generated
completely with AI.
So I I made the the script andthen I I cloned my own voice and
make myself do the entire umthe entire episode without even
speaking.
Okay.
So it was quite fun, but alsoscary at the same time.
Uh my husband couldn't tell thedifference between when it was
(16:09):
me and when it was the AI, butmy mom could.
Because I think moms they havea better understanding.
And this is something it'sfascinating, but I love the idea
that you can you have thisextra layer of protection.
Like if in in the case someoneclones your voice or your
likeness, you will be able toidentify it as something that is
(16:30):
not authorized by you.
Yes, and you can still do it ifyou want to.
Speaker 2 (16:41):
You you know you sign on digitally.
Uh but this is uh and this isuh uh an outcome of uh uh some
European projects because Iparticipated to to many European
projects as part of Universityof Geneva leading uh this kind
of projects, and some you knowEurope has not been very good at
uh creating unicorns uh so farfor different different aspects,
(17:03):
but uh at least uh for allthose projects, like hundreds of
millions uh uh uh that workedon this uh uh ID and signature
uh uh technology, now it's beingembedded in many countries,
even at the notary, uh anddisqualified electronic
signatures actually are anoutcome of some of these early
projects I made in uh 15 yearsago, or even like 20 years ago.
(17:26):
Uh so from an identity point ofview, this is good because it's
also uh embedded into laws aswell.
And even Switzerland isfollowing because um you know
it's it's the relationship withthe yeah.
Speaker 1 (17:38):
Oh no, that's amazing.
I really like that idea, and Ithink it's it's uh it's the a
good answer of every technologythat we're seeing uh popping up
and how easy it is to clone uhpeople and and how easy also to
be deceived by that.
Do you you think it's the realperson behind it, but it's
actually AI?
Speaker 5 (17:55):
Yeah.
Speaker 1 (17:56):
So um you work uh with artists, cities, and
technologists.
When designing system fortrust, how do you balance
(18:34):
technical verification withhuman intuition?
Speaker 2 (18:39):
So uh human intuition is is a bit dangerous from a
trust point of view because umall my work is interdisciplinary
work.
So my PhD was of course havinguh uh computational models, but
based on the human notion oftrust from uh social science.
Um so I um you know read manyarticles on uh uh what was the
(19:05):
the notion of trust in uh indifferent countries, different
cultures.
Um and from that you know Ibuilt a computational trust
system.
Uh then of course you you knowyou not only have uh trust
research in social challenge butalso economics, psychology, and
so on.
Uh uh I also uh integrated somemedical aspects because we did
(19:29):
some brain computing uh researchon you know what happens in
your in your brain when yousomeone trusts someone else.
Oh wow.
And if quite if a system knowsuh how to trick what uh what is
what tends to make a humantrusting someone else.
So, for example, from a facepoint of view, if you have a
(19:52):
very symmetrical face, uh youwill have uh the human in front
of you will uh have a tendencyto trust you more than someone
else who has a not a symmetricalface.
Uh so for example, even AI, forexample, AI could create faces
which are sim very uh perfectlysymmetrical.
Uh and so um some science isdangerous intuition because uh
(20:18):
the uh you know uh thousand ofyears, uh hundreds of thousands
of years of evolution uh that wehave had in uh trying to uh uh
As human uh trust someone elseuh could be used against you.
Uh you know, so intuition is uhis not really like uh what I
(20:39):
would recommend for uhtrustworthy uh systems.
Uh so then of course technicaltrust is much more important.
As we said, like for example,um I you know your intuition
would think that you know avideo or someone that you see uh
seems trustworthy, uh but maybethere would be no, as we said,
(21:03):
signatures, qualified electronicsignatures from uh as part of
this video from a real human.
And so then you would have analarm saying that be careful, uh
it's very trustworthy becauseactually your your smart glasses
could watch uh also uh wouldhave watched the the video uh or
(21:24):
the person in front of you uhand say, okay, uh yes, it seems
uh very trustworthy, but thereis no uh digital signatures uh
of a human uh behind it.
So be very uh be very careful.
Speaker 1 (21:36):
Well that's scary.
So um by the fact that uhthere's a science behind behind
how we react to certain to thestimulus, so how we react to
people talking, how their facelooks, and you can map that in
the brain, then you can createalso that with uh uh which you
could create an AI that emulatesthat.
That's that's very interesting.
(21:57):
So it's it's interesting.
Speaker 2 (22:00):
Interesting for records, yes.
Speaker 1 (22:01):
Yeah, interesting is scary.
So reality can really be uhmanipulated against us, yes,
against what the instinct thatwe normally would have.
Speaker 2 (22:10):
And from a social point of view, you know, usually
trust is not perfect because uhuh me, I really use trust uh as
a human notion of trust, not asa uh a security, which is
usually you know, in security,uh either you have the wrong the
right password or the wrongone, or if it is uh signed
correctly or wrongly.
But even signatures, maybe in10 years' time when you have a
(22:32):
quantum computing, the currentalgorithms uh uh could be
broken.
So a signature is never uhsure, yeah, it's sure for today,
uh, but maybe in 10 years' timeuh you know it will be broken,
so it will be you will needsomething else.
You really uh should uh becareful uh when you when you uh
(22:59):
deal with uh information infront of you and without
technical trust.
Speaker 1 (23:03):
Okay, okay, that's that's interesting and scary at
the same time.
So your work has uh touchedeverything from offline crypto
transactions to augmentedhumans.
What's the red threadconnecting all of it?
Speaker 2 (23:19):
Also, I mean uh it's a bit opportunistic, but um
anyway, uh my my my research isabout trust.
Um and uh and so of course whenthere is a new system uh
emerging and and created, uh I'mgoing to look at if it is it
trustworthy, uh how could youhack it or uh how could you
(23:41):
improve uh its trustworthiness.
Um however, uh myself uh youknow the uh because you know we
are we are moving from the uhuse of a smartphone uh looking
at screens to uh as you said ofthe more a metaverse where you
you would have would you wouldaccess uh digital information
(24:04):
anytime anywhere uh by soundwith like uh an AI uh on your
smart glasses giving you uh somerecommendations or some
warnings uh uh by what you see,uh but in in augmented reality.
Uh and myself I'm uh also intouh what we call like this uh
(24:25):
biohacking from a personal pointof view.
Speaker 4 (24:27):
Okay.
Speaker 2 (24:28):
Uh so uh from augmented reality to uh
augmented your human body, youknow, um in 2010 we started
these conferences calledaugmented uh augmented human uh
uh international conferences,uh, where we were the focus was
more on the human body uh thanthe augmented reality
environment.
Speaker 1 (24:48):
Oh okay, to enhance uh our capabilities.
Speaker 2 (24:51):
Yes, or enhance your capabilities uh or uh retrieve
your capabilities because wehave uh some tracks which are
about uh rehabilitation, youknow.
Someone, for example, wouldlose uh a leg, uh, and then with
an exoskeleton and you know,maybe uh some uh uh different uh
techniques, uh sensors in thespine and so on, uh, could uh
(25:12):
gain regain uh a leg.
So it's not only augmented uhit's sometimes it also uh just
gaining back your abilities.
Speaker 1 (25:23):
Yeah, of course, especially with people with uh
with this kind of disabilitiesthat by accident or a disease
they lost uh uh an arm, a leg,or or or something else, and
they they can have uh uh anotherchance of of uh regaining
mobility, full mobility.
Speaker 2 (25:39):
And so, yes, almost in humans, then uh I look at uh
some trust issues.
Um but then it's very boldbecause we not only have um like
digital systems, but also uh weuh look at also uh medicine uh
supplements uh and so on.
Speaker 1 (25:59):
Anything that can improve uh the human uh so it's
really augmented from the broadsense, it's not only the the
little part of technology thatwe imagine.
Okay, okay.
No, it's it's it's a great workwhen you when you can really
change people's life to for thebest.
Speaker 5 (26:14):
Yes, that's right.
Speaker 1 (26:17):
So moving to um online reputation, um, which is
something that have shifted verymuch and it's it I think it's
it's uh it keeps changing.
So it was once just aboutreviews, so uh five stars, two
stars, and so forth over peoplecommentaries in in certain uh
(26:38):
blogs or online.
Now online reputation can shapeyour credit, your identity, um
also your future employment manytimes.
Um, and even innovationfunding.
Where is it going next?
Where do you think the onlinereputation is gonna take us?
Speaker 2 (26:57):
Um yeah, I mean, so online reputation will be also
uh in the real world because, asI said, uh you are going to
interact to have access to thedigital information anytime
anywhere.
So uh when you're going uhsomeone is going to be in front
of you uh with your smartglasses, maybe uh your uh AI
will uh tell you a number ofthings about this person.
(27:20):
So uh any uh yeah, the problemis the reputation will be not
only accessed online but also inaugmented terms, like anytime.
Speaker 1 (27:31):
Uh exactly I will see you when I will see when I will
see you.
Speaker 2 (27:35):
Yes, I will see uh uh all the good things you did
because you you convicted it canbe uh on the good side as well.
Yeah, of course.
You can do uh uh when someonegoogles about you, it could be
uh negative, but it could bealso positive and find good
information if you made the theeffort, like with this blog, for
example.
Thank you.
(27:55):
Uh qua this vlog, or like Idon't quite or interviews as you
as you call it.
Um then uh maybe for your nextpromotion uh it will help you a
lot to get because you will benoticed, you know.
Uh so in a good way.
In a good way, yes.
So online reputation is a youknow uh sometimes you you see
the we see the bad sides, but ofcourse the the one which are
(28:19):
most popular, uh then they gainfrom it.
Uh and yeah, it will be in realtime anytime uh anywhere.
So it will not be online, itwill be augmented the
reputation.
Okay, yeah, augmented becauseyou you will retrieve anything
that you uh which is public oreven which was not public but
stolen and put on the dark web.
Speaker 1 (28:39):
Oh, okay.
Speaker 2 (28:40):
Yeah, so um yeah.
Speaker 1 (28:43):
So the the hidden files that someone got from you.
Speaker 5 (28:46):
Yes.
Speaker 1 (28:47):
Okay, but that's it is it is good in a way of sense
of security as well.
For for example, I'm uh I thinkit is into a in a social
setting when you're meetingsomeone for the first time, or
when you want to make sure thatit's not a clustered killer or
something like that, it's goodto know their reputation.
Speaker 2 (29:06):
Yeah, it's good to know.
Uh but um yeah, I mean, youknow, you you you already have
some some tools like PMIS, uh,which uh with only one image of
you can retrieve all the publicimages uh of you online.
Okay, and I can tell you uh,well, I mean, um uh when you do
(29:27):
something, you you you should beprepared, especially online,
even if you use a digital tool.
Uh prepared that sometimesomeday it could be public.
Okay, okay.
Yeah, uh and our generation,you know, we are not we didn't
we didn't know about the risksthat we were taking by by by uh
(29:49):
you know all these generations.
Uh they don't were on TikTok,for example, when they publish
something, yeah, this is thereforever, it's been collected
already, yeah, and could beretrieved back.
Uh uh, and even uh yes, I meanuh even if you changed uh and
you have some TikTok videoswhere you were uh maybe drunk uh
yes, uh this will be uhretrieved uh and be seen by uh
(30:19):
in 10 years time.
Uh you will have uh so so youshould be prepared that uh I
anyway, even if you don't wantto do it digitally, you have so
many sensors everywhere uh thatuh we should be prepared to be
to accept that uh we are beingwatched, we are being watched
(30:41):
the full time.
And so uh well I mean you it'sdifficult sometimes uh to to to
restrain yourself or to yeah ofcourse but um uh yeah some but
then then uh yeah yeah and thenit will be very difficult to to
to protect it.
Speaker 1 (30:56):
So yeah, it's um it's it's interesting in that way
that uh there's a lot of um umbelief that privacy itself is
not gonna be something that isuh the same sense that we have
it right now is not gonna existanymore.
Because we will be constantlybeing observed, either either
monitored by devices that arearound us or by the the the ones
(31:19):
that are outside in the publicwhen we go outside.
So there's never gonna be apure privacy moment um in the
near future.
Speaker 2 (31:27):
Yes, it's it will be possible uh to remain private,
uh, but uh it will be uhdifficult uh and this is
difficult to remain private fora long time.
Speaker 1 (31:39):
Yeah, yeah.
Well, interesting.
So you explored uh the legalweight of digital signatures on
NFTs.
Do you believe we're movingtowards enforceable trust in the
metaverse?
Speaker 2 (31:53):
Although uh trust as a human notion of trust uh is
not enforceable.
Um this maybe you could have uhuh a number between zero and
one hundred.
Um but you see you cannot forcethe whole world to trust you.
So you see, from a world pointof view, you cannot force
(32:15):
everybody uh because the worldis uh fortunately is still uh
has still some independence.
Yeah, uh so uh from a socialbecause trust uh is also linked
to your social network.
Uh from the sociology point ofview, uh we always talk about
those social networks, theyexisted before, uh of course,
when the when you know uh anumber of uh you have your
(32:39):
parents, your family, yourfriends, and then they know
other people, uh you you havethis kind of uh social networks,
they existed for a long time.
Uh now with the social networksyou have maybe weaker uh links
because uh, for example, onInstagram, so the new generation
we are doing some research atthe moment at University of
Geneva.
Uh before it was more like thethe brain is more made to really
(33:01):
have close quite some a circleof uh 150 people.
But uh it's not uncommon thatuh the younger generations from
15 to 30 years old now, uh theyhave uh 500 uh for the men to
1000 uh weak followers or orcontacts, you know, on those
(33:24):
Instagrams or social networks.
Uh so it's getting uh it's it'sgetting broader.
And anyway, uh by you know thetrust also involves
recommendations from others.
Um so when you build a trustmechanism when reviews, for
example, okay, you could havereviews which are non-linked to
(33:46):
you, but if they are if youryour close friends uh also
provide uh reviews, uh andsometimes at some some stage
they they try to add this intouh uh into triple advisor.
For example, you could see uhyou could connect with your
Facebook account, you know youhad a Facebook Connect, but you
know now everybody uh is afraidof Facebook Connect.
(34:06):
So but at some stage it worked,and then you were seeing uh
reviews of hotels of yourfriends.
Speaker 4 (34:11):
Oh, okay.
Speaker 2 (34:12):
And so uh somehow, of course, uh you tend to trust
more the people, the closefriends than than uh than others
if you uh if you if if you knowthat they are uh stable.
Speaker 1 (34:28):
Yeah, because it it it makes more reference to the
same person, so moreconfirmation about who the
person is and how it be theybehave.
Speaker 2 (34:36):
So you see, you cannot enforce.
Uh you can try to influencebecause you can try to build on
your social networks.
You can as a hacker, you cantry to uh uh become the friend
because if someone is very uhhard into uh getting into uh you
could try to select to toanalyze the social network of
the uh of that person and try tosee uh well position uh
(34:59):
profiles which seems to be easyto to trick into trusting you.
And because then after youbecome a friend of this person,
uh and someone, for example, onLinkedIn, you know, is trying to
uh a hacker is trying to getconnected into your so linked uh
social networks, professionalnetworks, but it's a social
(35:21):
networks, uh you would try to uhbe already a contact of someone
you know, and you can see youknow, oh this guy knows this
person.
Uh and uh by that, if there isno double check, uh you would
see a request from this uh otherperson or contact of uh a
professional contact you know,so you it will facilitate this
(35:47):
request uh more easily than thanum if uh it was there was no
connection.
Speaker 1 (35:53):
You have no one in common.
So how how did he get to me ifI have no one in common?
Speaker 2 (35:57):
Yes.
So uh and then you cannot conyou cannot have a social
contract network of the wholeworld.
That's the same.
Uh uh the farthest you go, uhthe less influence they have.
Um but then you you you can tryto influence, uh, that's what I
say in my courses.
Uh, online reputation at theMedia Lab in uh University of
Geneva.
We have a uh online reputationmonitoring um course, and uh so
(36:22):
it's not enforceable, it's um uhyou can influence it, but not
enforce it.
Speaker 1 (36:26):
Okay, interesting.
I would I would like to knowwhat it comes out if you put my
name.
A lot of things about IPprobably.
So with the rise of generativeAI, which is in everyone's mind
right now, um, what's your takeon trust boundaries?
How much should we reveal aboutwhat's AI made?
(36:49):
So if I generate uh somethingwith AI, should in every every
moment I should always declarethat I made it with AI, or can I
just make some things andwithout saying that it was AI
generated?
And how would that affect thetrust?
So the people will expect thatfull transparency.
I would like to know everythingthat is AI generated, I would
(37:11):
like to know beforehand, or itcould it could be more flexible,
like maybe not I don't need toknow everything, just things
aren't sensitive, or that canaffect me.
Speaker 2 (37:21):
The thing is uh AI is going to be everywhere anytime,
anytime any anywhere.
Uh you know, as I said withaudio or even like uh in in in
other terms.
So um you are it can generate,but it can also recommend you
things, so uh it will inf itwill influence uh anything you
(37:43):
decide.
Um so to me, I mean it's morelike um we are more um we are
more interested in in knowinghow much human work has been put
into a product or uh art or uhactivity uh than knowing that AI
(38:03):
was used because uh in 10years' time uh AI will have been
used for everything almost.
Speaker 1 (38:11):
Okay, it's gonna be the norm.
Speaker 2 (38:13):
So you would have a warning for uh for everything
constant everything.
So uh but then uh on the otherhand, uh how much human work has
been uh done for a specific uhproduct or task uh is uh more
interesting to me.
We are working on uh as part inGeneva as part of a project
called House of Good AI.
(38:34):
Uh, we have also a good AI uhuh society uh when we do some
meetups, and we discussed um uhthat it would be interesting not
having a uh human-made labelbecause uh if you craft you
craft uh uh like for wood, uh ofcourse it could be be 100%
(38:55):
human-made.
But if it's a document or ifit's anything else, um uh you
you you would use already uh acomputer which is not human-made
or uh or some some someprograms which are not
human-made because uh some AIhas already uh participated to
it.
So um it would be uh moreinteresting to uh than to have
(39:21):
uh 100% human-made, to have uhuh and what what we call uh uh
in house of good AI we we havethis way of uh uh allowing any
any people uh to uh have theirown cryptocurrency, which one uh
cryptocurrency of Jean-Marc isequivalent to one hour of my
(39:42):
time, and you would have uh uhyour own token uh uh which is uh
one hour of your time that youhave lived, and uh you could
decide uh how much of thosetokens are are uh being spent
for some activities.
So, for example, in art, uh apainter could say, Okay, uh this
(40:03):
uh painting I made four hours,uh it took four hours of my
time, but also an AI generatedvideo, even if you uh used uh AI
generation, it it took some ofit of your time to uh to to
spend.
So you could say, Okay, it tookme 10 minutes to uh to write
the prompt and so on.
Uh and then uh instead ofhaving a percentage, because
(40:24):
it's a percentage is difficultto really um uh compute or to
have the right information uh bycounting the number of tokens
that the product has uh has beenspent for for this um then you
would have an estimation uhestimation of uh uh how many
hours of work of human work hasbeen spent uh to this product or
(40:47):
even like just paid asemployment uh sometimes we we we
we are laughing because um uhyou know some people they say
that they have a bullshit uh jobuh meaning that you know they
they are not doing much for thecompany but uh I mean they are
here and they are paid uh so uhinstead of having a uh a minimum
(41:08):
uh salary that uh uh tocompensate that you don't have a
job and um and you are onlypaid uh an amount maybe an AI
could pay you uh just to spendsome of your time tokens for uh
uh for having a a per like anumber of uh human work paid uh
(41:29):
even if you did nothing but atleast uh uh maybe close to your
your your social network to givesome uh value uh other value uh
human value to to the productthat or service that the the uh
they buy so you're thinkingabout that this uh very long
debate about the the generalminimum wage that people are
(41:52):
talking about that everyoneshould receive a minimum wage if
even if they're not uhespecially those that are not um
able to work just yet or theyoung people that uh still don't
are not enter the workforce uminstead of having that just
because of existing then theidea would be if you are putting
yourself into um work as ahuman hour for for AI purposes
(42:17):
would be a good way to earn thisyeah an AI could could choose
you know uh like some people uhit's kind of hiring but maybe
doing nothing it could be moreflexible yeah could be do could
be used to uh maybe uh marketthe the product to some to their
social uh circles okay uhbecause maybe the AI has spotted
that uh uh this is uh this is awealthy uh family uh um and so
(42:46):
uh yeah in this way I mean atleast uh it's a bit clearer I
mean also well positioned peoplecould get more uh paid more
could be paid more for that yeahinteresting so so it's uh
there's a lot of uh there's alot of ideas here to um to
address that this um the oneabout most things will be AI
(43:08):
generated anyways in the nearfuture so the labeling should be
for human input not for AI yeahuh uh uh uh you would have uh
on the label like uh the numberof hours that have been uh spent
and proven you know that thisis secure on uh the the House of
Good AI have a we have a um uhan an attack resistant uh
(43:29):
algorithm uh also put on apublic blockchain yeah uh and I
mean cheating is uh uh we'remaking mechanisms uh against
cheating so it would really uhkind of uh it would be public
information uh that any uh anyAI could uh retrieve on on uh on
(43:50):
the blockchain.
Okay interesting and moving onto the next question how do we
prevent algorithm trust frombecoming automated bias I know
(44:16):
um as I said my vision of trustis more like a human notion of
trust and uh uh it will tend tobe biased of uh based on your
social network and and historyand so on uh and trust is not
global as I said uh twodifferent persons could have a
(44:38):
very different trust value insome something else due to their
social network and and pastexperience and recommendations
uh and anything that that thathappened so um um yeah um so
there's there's always gonna beuh bias and this it's yeah I
(44:59):
mean to me it's by by design itis biased okay because we are by
design yes but on the other onethe uh the bias uh will uh uh
tend to uh to be uh to makedecisions or um to to be qu
(45:20):
would would be uh used uh andand and should the the goal of
of this system or this trust uhmechanism is to have the the the
most the optimal risk rewardyou know so um usually I mean
the bias is the ideally uh ifyou have good trust mechanism uh
(45:41):
metric mechanism uh would um bein your favor because it could
be biased against you um so itwould yeah yeah so we it it's
something that is uh is humannature to to have uh different
judgment and to have differentapproaches yeah uh if if you
take trust as a human notion oftrust uh it is biased uh you
(46:05):
would you you will tend toprotect more your family than uh
than strangers in many culturesalmost all cultures because uh
you know the maybe the ones thatwere not protecting their
families they they disappearedalready due to evolution makes
sense if you could embed onetrust metric into everyday
(46:26):
platforms what would it measurewhat what it would measure um to
me I mean it it would be uhmaybe based on one of my early
(46:58):
work uh that I uh designed uh itwas called the trusto um and
really is the the would be thehuman trust you know the if you
have a human component commoncomponent in your system would
be uh the human uh uh trust uhof the current users you know
(47:22):
the the trust of the of thishuman uh regarding the mechanism
or system uh which is being uhevaluated uh so it would be the
human notion of trust.
Okay okay then what triggers usto trust uh I mean as as the
(47:43):
component in in the distrustmetric okay uh it would be
important to take into accountperfect so now we have the the
fun part of the episode um Iwill ask you um one um I I'll
(48:07):
ask you to pick one of these twostatements so first one is
verify synonyms or totaltransparency or verify you mean
like uh uh that it there there'sstill a synonym you don't know
exactly who is behind but it'sverified that it's a person but
(48:29):
you don't know the name of theperson then the second one is
total transparency of course uhcome to me at least verified so
it's gonna be uh um flower oneto three if you know it's a
verified user but you don't knowthe name of the person okay
perfect NFT has proof ofownership or traditional IP
(48:51):
registration uh NFT has proof ofownership maybe sign NFTs you
know one global language poweredby translationships or keep
every language alive I mean keepevery language alive but uh as
we said uh AI will translateeverything uh very easily so uh
(49:14):
so a combination yeah I mean nobut uh it's it's very
interesting I think to have uhthe diverse uh diversity into in
into in the languages um butyou know AI will translate uh
you know automatically what wehear so um but then maybe it
(49:35):
even like somehow it might itmight protect maybe more like uh
uh less spoken languagesbecause uh usually what happens
is that you need to learn thenational language to to be able
to work in in the nation um butthen if you know uh at that
(49:59):
stage if when you speak yournative language or like local
language uh dialect uh it'sautomatically understood by the
person in front then uh you haveless uh you you need less to uh
to to forget about your dialectand and can keep it it will
keep them alive because then youwill be able to use it really
(50:19):
without uh um without withouthaving to go to and learn the
other one yes so I mean maybethat that's the other way I mean
maybe it's about it's positive.
I love that I love that so touse the AI has a tool to
preserve the language yeah sodecentralized data governance or
citizen led data trust sopeople will run the data or a
(50:42):
decentralized way to run thedata the the citizen would um
would own their own data or itwould be centralized yes uh no
it should be uh you should quiteideally your privacy and your
private your your data shouldyou should uh uh be able to own
(51:05):
it and decide when you uh wouldhave to share it uh yes I'm more
in in this favor like uh forfor this than than uh of course
having a centralized uhmonitoring system yes so citizen
led immutable records ofeverything or selective
(51:27):
forgetfulness by design do youhave the right to be forgotten
or not um but it depends I meanum you mean after you you you
you uh you die yes but um umdifficult to to to uh to uh to
(51:52):
answer us those questions yeahthat's that's the idea because
they both of them have merit yesum I think for like from a
freedom point of view uh onceyou are died you you you should
have the the right to be uh tobe forgotten because the the
(52:14):
thing is that if you didmalicious activities uh and you
are still alive and they areforgotten uh well I mean uh you
cannot be uh judged uh for forwhat you uh what you did but
once you are you are dead so Ithink uh yes I mean uh
forgetting uh choosing to thatyour data is erased after you
(52:35):
you die uh I think for me wouldbe fine before uh especially for
criminal activities yeah ofcourse those should be kept
because there's public interestfor them yeah and then they were
they are they were malicious uhyeah yeah okay so a combination
of both transparent uhtransparent governance or
(52:57):
invisible protocols that workwell it depends for what also uh
uh so some system you like uhtransparency for example maybe
for governments it would be niceto see exactly uh what you paid
for how they reached thatdecision how they how much they
invested in that uh did it go tocorruption or not how much went
(53:20):
to corruption okay uh but um uhyeah so it depends on on on
which system you are you're youare talking about for the public
uh goods uh yes I mean thegovernance should be transparent
uh when it matters of uhprivate uh uh aspects uh as I
said ideally privacy uh shouldremain but is it possible yeah
(53:43):
okay AI trust ratings issued bypeers or issue by regulators uh
could you repeat the first onesorry I mean uh so AI trust
ratings yes either issue bypeers or issue by regulators do
you trust more what the peoplethink in general the general
(54:05):
public think or what a regulatoruh decides that if if it's
truthworthy or not uh but thenwhy why do we say AI trust?
Because uh so it's a let's sayit's an AI tool and they're
ranked the option is is eitherranked uh by the public or it's
ranked by the regulators by thegovernment who would you trust
(54:27):
more to rank the AL2 you thetool you need a mix of both
because um uh in many uh in manyaspects uh you have a
competition between theattackers and the defenders and
your government uh as much moneythan uh uh than yourself
usually uh so they can go intomore details uh about some
(54:51):
technical aspects um and again Ithink it we need both yeah so
good so every everyone will havetheir own role on on assessing
the tool yeah you you you youyou need to have your own
thinking uh of course uh butthen as I said uh sometimes um
you wouldn't have enough to uhto assess you know the what is
(55:13):
in front of you and so uh thenbecause it's your government uh
I mean if you then you also needhave to trust your government
of course but uh usually I meanuh in democracy it's uh it
should be uh in your favor uh toum then you need the more okay
uh bigger uh players to toassess what is in front of you
(55:33):
uh to spend a lot of resourcesand and knowledge uh and somehow
uh because it's your governmentuh you tend to give it a bit
more trust than uh if it was thecompeting government you know
okay makes sense ecosystemswhere trust can be restored or
(55:55):
ones where it is lost it isirreversible no you can you can
always uh uh try to recovertrust but this is uh difficult
you can lose trust you know ittakes a long time to build trust
in the human uh world um ittakes more time I mean to build
(56:15):
trust than to lose it yeah it'strue you can lose it in one
second you can upload one memoryto the internet private forever
or public for everyone um couldyou repeat again this is a bit
(56:36):
uh sci-fi kind of thing you canupload one memory to the
internet one you mean one memoryso one memory from your life to
the internet and you willallload it to be private or to
be public no but of course Imean the the same uh depends on
uh it should be good to thatit's it's kept uh as a as a as a
(56:58):
souvenir uh memory uh but thenyou should be able to to choose
if it's private or not I meanthen again as I said now as soon
as you put something online uheven on a supposingly secure uh
service uh when it's digitalizeduh uh it can become public so
um but ideally uh you should yesthe for me I'm uh for the human
(57:24):
privacy and so uh should beable to choose uh uh if it's
private or not.
Speaker 1 (57:30):
Okay, so for everyone's to choose.
Speaker 2 (57:33):
Yes.
Speaker 1 (57:34):
Okay now uh take the palette okay and this is this is
part of the game of true orfuturist.
Okay.
So I will give you a statementand you choose that is true that
is meaning that it's happeningright now or futuristic that is
going to happen in a hundredyears or it's never going to
happen.
Okay ready?
(57:55):
Yes NFTs will be legallyequivalent to a notarial
signature so like a notary umyeah also then uh true true
augmented humans will carrydynamic trust layers visible
through augmented reality so wewhat we were talking about
(58:19):
futuristic is what like uh fiveyears ten years or more no no uh
50 years uh more more morefurther in the future uh true
true okay because we were justtalking about that before more
in ten years time okay ten yearsokay interesting web3 platforms
will have ethics protocol votedon on by users um yeah I think
(58:48):
true true interoperable identitywill become a requirement for
cross platform login no morepasswords and usernames um yes I
mean the interoperable uh uh II access this is true also you
(59:10):
know no problem yeah so before50 years of course before 50
years I I would love to be ableto forget all the usernames and
passwords that I have to createevery time but then you might
lose some privacy yeah that'strue but you can never uh
there's uh my husband says thathe doesn't believe in progress
because whenever you gainsomething you lose something
else so there's always gonna bea bargain you always got there's
(59:33):
always gonna be a bargain soit's interesting because it's
true it's that you you have toyou don't have to hurl out of
anymore of creating usernamesand passwords but then you have
the issue that someone can takeit and then you're you're locked
out of everything.
Speaker 2 (59:48):
Okay conflict resolution will shift from legal
courts to smart trust ecosystemI mean the lawyers are are very
uh strong even if AI is isgetting uh smarter um so I think
(01:00:08):
uh I may think in future westill need uh uh lawyers okay
that's so so futuristic it'sgood so we still we still need
it okay your rowsing behaviorwill be used to compute
trustworthiness by insurancecompanies
Speaker 1 (01:00:31):
Well this is this is uh ah well usually the is it
really important for them to toto know uh your trust or sinus
is more like uh your elf no orif if they from that they can
get real information that ifyou're googling uh something
that is uh I mean or you'rebehaving recklessly.
Speaker 2 (01:00:54):
I mean uh if they if they succeed uh it's the it's a
bit difficult.
Um but um yeah more and it'seasier and easier uh uh as I
said because we have more andmore information on online uh or
digital.
Speaker 1 (01:01:10):
Okay.
AI will have to explain itselfbefore making a decision about
you.
Speaker 2 (01:01:24):
In 10 years' time, I mean of course uh it will be
already here.
I mean they will uh help us toto make decisions, so I mean
it's true.
Um but sometimes I mean you youyou know uh people they when
they they they have to to makesecurity decisions for example
uh as we have seen for cookieson on website and so on, usually
(01:01:46):
I mean uh they don't have timeto to read the the the licenses,
uh they don't have theknowledge, for example, uh uh or
they don't have the money uh topay for some uh strong AIs or
mechanisms to help them.
Um and so because they don'thave time, uh for some decisions
(01:02:07):
the AI, I guess uh people willuh clearly let uh AI deciding
deciding for them because uhthey don't have time and they
are sometimes maybe also uh lazysometimes and then of course
there are risks and then maybeuh but then explaining
themselves they they will beable to um to do that in 10
(01:02:28):
years' time.
Speaker 1 (01:02:31):
We'll need a license to use powerful AI tools like we
do with cars.
With cars just or with amedical license, for example,
engineering uh um certification,so uh AI tools that can really
create something that could posea harm to the public.
(01:02:53):
Would that require also like acertification, a license, or an
official authorization to usethem?
Speaker 2 (01:03:01):
Well, I think it's it's true because that that's
what they are trying to do now,but uh if AI continues to to
make progress uh like that, uhin 10 years' time or 20 years'
time, we might have uh superintelligence.
Uh uh and so uh this will be abit uh you know uh we won't
(01:03:23):
understand you know what what'sgoing on, and and so uh it's
gonna be beyond us.
I mean um so um now it's acase, but then it might
disappear.
Speaker 1 (01:03:35):
Okay.
You receive personalizedpersonalized laws based on your
online behavior.
Personalized laws that applyonly to you.
Speaker 2 (01:03:50):
Well, yeah, you know, I'm not uh I'm not uh uh into
legal uh like nanny state kindof thing.
Uh no, I uh no uh uh well Iguess law uh should be a bit uh
like for a country uh size orlike uh it's uh uh a kind of a
uh regulation uh uh uh for acommunity of people.
(01:04:11):
So only for you.
I don't see it.
So I mean uh because yeah, youknow it's not even now, not in
the future, it's never no, neverso yeah, it's never like this.
Speaker 1 (01:04:24):
Final question.
You'll take person you willtake personality test before
being allowed to use certainplatforms to see if you're
messed at a stability or youhave the connectivity uh or you
have the development.
Speaker 2 (01:04:40):
I think this is already known.
I mean, uh uh, but in thefuture, as I said, we we won't
use platforms.
Uh we will interact with themetaverse like uh uh
automatically, like everydaylife.
Yeah, anytime anywhere.
It's kind of uh what we callVaser, for example, uh was uh
calling it uh ubiquitouscomputing.
Uh and we already worked on onthis, like a global global
(01:05:03):
computing, you be ubiquitouscomputing.
He was talking of thedisappearing computer.
Um yeah, and then everythingwill we we we'll we won't have
platforms anymore.
Speaker 1 (01:05:14):
Um so it's gonna be the everyday life?
Speaker 2 (01:05:16):
Yes.
Speaker 1 (01:05:17):
Okay.
Is this so true?
Speaker 5 (01:05:21):
Uh yes.
Speaker 1 (01:05:26):
Okay, so now um the question to wrap the episode.
Has technology gets faster andmore visible?
Was one human truth you believeevery trust system should
protect?
Speaker 3 (01:05:47):
Human truth.
Speaker 2 (01:05:49):
Um, this is very narrow one, but uh let's say uh
I would say uh freedom.
Speaker 1 (01:06:01):
Yeah, I love that.
It's it's the it's the core ofhumanity.
Thank you so much, and Marcushas been an amazing
conversation.
Speaker 2 (01:06:12):
Sometimes difficult, but uh but very well thought,
you know, like you you spent alot of time to to to think about
these uh these concepts andeven yourself.
I mean you you uh you have avery good you did a very good
brainstorming, so it wasenjoyable.
Speaker 1 (01:06:25):
Yeah, that's very sweet, thank you so much.
No, I had a great time.
It is it was very, veryinteresting, scary in a good
way.
Because the future is is uh issomething that is um enticing,
and we always want to know whatis going to happen next, but it
can be also scary if uh if thepeople the right people are not
(01:06:47):
making the right decisions.
So yeah.
Yeah.
So thank you so much forreminding us uh that trust is a
framework, and in a digitalworld, that framework must be
designed, debated, and defended.
Your work shows that what wetrust shapes what we build.
And if we want technologiesthat empower, not just impress,
(01:07:09):
it starts with a reputation thatis earned, not bought.
Because the future isn't justdecentralized, it is deeply
personal.
Speaker 2 (01:07:18):
Thank you.
Speaker 1 (01:07:19):
Thank you.
Speaker (01:07:24):
Thank you for listening to Intangibilia, the podcast of
Intangible Law.
Plain talk about intellectualproperty.
Did you like what we talkedtoday?
Please share with your network.
Do you want to learn more aboutintellectual property?
Subscribe now on your favoritepodcast player.
Follow us on Instagram,Facebook, LinkedIn, and Twitter.
(01:07:44):
Visit our websitewww.intangibia.com.
Copyright Leticia Caminero2020.
All rights reserved.
This podcast is provided forinformation purposes only.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!