Subscribe
Share
Share
Embed
Join us on the Disruption Now podcast as we challenge the status quo and advocate for digital equity, ownership, and responsible technology.
welcome to disruption now
I'm your host and moderator
Rob Richardson with me is Richard Hartnett
he is the director of policy for cyber
at the university of Cincinnati
he's a man of many many talents
he brings together policymakers
academia innovators and researchers
and trust me those are a lot of different bureaucracies
to bring together so he's a man of many talents
it's a lot of work
and we're talking about a subject that
should keep you up at night
we're talking about cyber security
and so if you wanna know the threats that are out there
why you should care why it matters for America
why it matters no matter who you are
this is your episode we might scare you a little bit
but being being aware is better than being unaware
so that's our hope that that you gain from this
as always I like to say we have about 900 followers
we have many more people that watch
and listen to the show
in order for us to take the full message of disruption
we ask if you're watching us here on YouTube
take the time to please subscribe
even comment every now and then
that helps us keep the disruption
and disrupting for good so
more people can be empowered to learn about technology
we hope to see you and as always
keep disrupting so with me
Richard Hartnett Richard
it's good to have you on the show brother
it's great to have you be here with you Rob
but I think you just eliminated my cocktail party
invites you know like hey
this guy's just gonna scare the hell out of you
don't invite him but no
I I totally agree with you you know
these are
these are important topics that affect us all
and so thanks
thanks for giving me a chance to chat with you
all right
so you've you started off at the which college
the College College of Arts and Sciences
walk me through how a man that starts off at the
college of Arts and Sciences
walks his way into cyber security and policy
how did it how did how did that story happen in yeah
I know it's a it it
it is actually an interesting segue
a point in this story
where I find myself down at Georgia Tech
at a conference
and it was about 150 computer scientists
engineers and I'm like
literally the only social scientist in the room
to the point where the organizers at the end said um
can you get up on stage
and just explain what the hell you're doing here right
so that's how and we're talking early nineties here
the intersection of thinking about cybersecurity
and social science strategic studies policy
it just wasn't happening and it's
it's been a slow roll for that
but my entry into it was actually uh
before probably a lot of your uh listeners uh
were alive I were born
late 1980s early 90s
my work is focused on nuclear weapons and how
how we deal with deterrence
and all of that type of stuff
and I came up with a
a different model of thinking about it
that's a story for another podcast
but did a little work for the department of defense
and they said hey
you know it's pretty interesting
we got this thing called a browser
not sure whether it's gonna become a big thing uh
but if it did
you know what's the implications for security
sure so I
I I kind of applied the
the work that I was doing and came back and said yeah
we're
we're in trouble like the way we think about security
the way we've been thinking about it for 70 years
doesn't map to a network competing environment
I didn't have the answer to what the solution was
but at that point to use your term
I saw disruption and we were gonna have to disrupt
the way we thought about security
how we organized for security
and the capabilities that we brought to the table
the problem was I'm saying this in the very early 90s
and then everybody
started making money on the internet right
so we have the dot com sure you know
boom and myself and you know
just very few other people are like yeah
that dooming gloom kind of thing you know
like this is this is okay
but this is gonna challenge us on the security side
absolutely
and you were right about that as it turned out yeah
you were right but and and I and I say this Richard
cause this is a good intersection
so you mentioned a few things that
I actually think
have parallel to what's happening right now
in this current moment that we find ourselves in here
in 2025 we had to redesign and re
and re imagine and rethink how
what security look like when it come
when it came to cyber security um
because we were moving into this online presence
that we moved into
so you brought up the 90s
and I always like to say history doesn't repeat itself
but it rhymes so when I look at where we are
you brought up looking at reimagining security
you start off with nuclear and I have a
I have a plug for that too
but you also talked about how we have to reorient
ourselves
reimagine what security looks like across the board
and of course
you were right I feel we're at a similar moment uh
that could be even more exponential
I've often said the
next five years are gonna
be more disruptive than the last 50 given uh
generative AI given likely quantum computing
what's happening here so thinking about that
what learning from what you see in the past
what do you see as the biggest opportunity for threats
and what should the pivot in our minds
be like
to be prepared for as individuals and as organizations
as we look at AI and cybersecurity yeah
no that's I think you're absolutely right Rob
there's there's these times in history right
that we have in in history books
the Dark Ages and that gets like
now you're really going to your professor bag
yeah I am
I am but you know
it'll get you know
a page right it's a couple hundred years
yeah right
people were doing lots of things
they were living on this planet
but it wasn't an interesting time
it wasn't a time that actually changed the trajectory
so it just gets a page in a history book
but then you open up a history book
in Industrial Revolution that's a whole chapter
yeah because that created a massive socio economic
political shift in the way we uh
organize society and I'm in the camp that suggests
that this next technological um
step function it's not incremental
we're gonna step function
off of algorithms and computers
describe what a step function is for the so layman
you know technology continues to move it
it improves typically incrementally
but there are these times where you get this
confluence of lots of different factors
and in this case it's the math that's behind algorithms
and massive compute power
in the form of these new people here
about the AI chips and things of this nature
we're bringing those things together
and they are going to give us a leap
a big step up in what we are able to throw at it
in terms of complexity yes
and so we have really big hard problems in society
whether it's on health whether it's on the way people
uh should be educated how
how better we can improve education
what are new job opportunities
etcetera etcetera
we're going to be in a
in a state where we can take massive
massive amounts of information
that we've accumulated throughout human history
yes and bring that together and gain insights at scale
right the scale and scope of information processing
we you know
humans are like really good information processors
like it's an amazing thing
our brains
but we have a limitation of how much we can consume
in a 24 hour period
these algorithms don't have a consumption problem no
they they don't get a full right no
you know they can just turn and turn
and turn and turn at tremendous speed
so we can get scale and scope
and actually not have to wait years
decades centuries for big
big change so that's gonna
we're gonna be in a period of really compressed change
we've always had change in human history
this is gonna be a period of compressed change
so the industrial revolution
big dislocation big disruption
but it took some time yeah
so we had time to absorb it
and I think what's the most fascinating
most the biggest challenge we're gonna face
is the fact that we're gonna have to deal with change
in a very compressed time frame
yes and I think
so I wanna I wanna dive into that
I like to dive into the security concerns yeah
um
but I wanna take it back a little bit
like I feel
one of the biggest problems we have when it comes to
algorithms is not new uh huh
it's the fact that we have a
the algorithms are really
creating perceptions of people's mind
of reality right
through social media and other things
I feel what I feel
one of our biggest challenges isn't
necessarily those things that you said
not that they're not important
I feel it's our ability to agree as society
about what facts are well right
and so yeah
so I know I I agree with you
so so I have a question
yeah I have a question
so when when we can't like
what do we do with the threat of AI there when it
we know it has the ability to tell us that we are
right about everything when we could be wrong uh
how do we
how does that also inform how we think about security
cause I see multiple kind of paths there yeah
so so I
I I'll
I'll tie what I was saying OK
to that question and I think we actually
there's not much space between us OK because
the way you develop an understanding of some complex
problem or something that's right in front of you
is to reflect on it yeah
so when we have shades of truth
the thing that enables
um us to be able to pull ideas out from
a lot of falsehoods yes
typically has been time to reflect on it
hmm time to bring evidence to bear
yes to bring expertise to bear right
so the reason I went to this fundamental issue
about compressed time is it
I think it explains
a lot of the challenges that we're going to see
make sense OK
OK so you've just identified a critical one correct
and that is how do you manage society when
truth if you will
or at least consensus yes
is is so fluid to
to the point where we can't latch onto something
and the algorithmic space that we're in
whether it's on the social media platforms
whether it's in our workplace
where we're
we're going to a large language model like chat GPT
right and saying
you know what can I get this information
oh wow
that's a great summary
right I'll just use the
and so that has another effect on your problem
if we don't reflect we can't
take falsehood and and truth and compare them right
without reflection right
but how do we actually what's
what's behind these things
you have to have a little deep thinking
these things don't encourage no deep thinking right
encourage action
tech right now is about convenience
and it's convenience to the point at which
it's putting decision makers in a really
really tight box information box
so we've we've heard about this
you know phenomenon of being in a
in a point where I you know
I go to my friends group and
and all I hear are things that I want to hear
so that reinforce whatever I want to hear yeah
people talk about echo chamber right
I think it's a deeper problem
and this gets to your
your issue is that not only do we not have reflection
but we don't actually have critique nope
we don't actually have the time to discern
well wait a second
you know that
that's just not that's not solid
right that
that that's not true
so the large language models is first iteration of
of of algorithmic um
impact yeah
I've been teaching for 34 years right I got one maxim
students are like water
they'll go to the point of least resistance
yes right
so just like people yeah
that's people
so we have to structure education to be challenging
yes
and that challenge has to say it's okay to be critical
not hey
you're wrong right
but reflective of saying how did you come to that
how did you come to that conclusion
right and ask that question
if we get really really good summaries of we
we allow the AI
to summarize a whole bunch of different articles right
let's just assume it does
a good job like
there's whole problems of hallucination
and not getting it right
but let's even assume that it's a good summary
if I don't go beyond the summary in my thinking
then your problem becomes even
you know more exacerbated
right because now I'm losing the skill yeah
that's true but it feels like Richard
to be critical Richard
it feels like that's happening some right oh
I I think that's the path we're on unless we arrest it
so that's that's my question
what does a rest in it look like
what does that look like for education
what's the framework for approaching policy
on something like this
so we right now um
so we don't have a strategy
we don't have a a tech strategy in this country
I don't think there's
most countries don't have a tech strategy
China does um
they do but it's a different form of government yes
but but
but it's a fact but so let me
I'll rephrase but so
so I don't think any democracies have a great OK
yes tech because we have and
and there's good reasons to use the market
the free market to be an engine of innovation right
absolutely so you don't want to box off possibilities
that's great but there's no directionality to this
I agree so when we talk about strategy
strategy is
anchor it on a place you wanna be at the end
and then you you figure out the resources
and means by which you wanna get there
I completely agree and right now
what we're doing is this free for all
bottom up approach
there's gotta be a point at which we say
we've got a good base here of what
algorithmic agency is gonna look like
in the next five years
let's take a point to reflect on this right
and say
what direction do we want to go for the American people
that's exactly right do we want to make life so
and you can have different objectives
yes right
most people want to have um
more wealth yes
well there's a pathway by which we leverage this
for more wealth but that might um
but wealth for whom
yes right
which is a which is a sociopolitical question right
and if right now
it's wealth for people that hold stock
in the companies and own the companies
it's not clear right now that we have a policy of
that's gonna drive us towards uh work yeah right
so if people don't have jobs
and therefore don't have income at the macro level
then this thing is not a wealth generator for society
it's a wealth generator
for a particular part of society yeah
but that's a that's a policy decision
that we have to start to have a conversation about
there's tax policy that can drive this
there's regulatory policy that can drive it
and if I can Richard I
I also think there is
there are a lot of other questions too
the economic one is something people
obviously think about and it's very central
and I agree with all your sentiment
that we have to think about
you know what type of society we
we want to build
we have to do this with the industrial revolution
mm hmm and um
you know looking at history
I mean we eventually got it right
but not not without a lot of mistakes
yeah right
I mean
we went through two world wars we went through um
you know there was literally uh
there were people were saying well
maybe the German model was the better model when
when when Nazism was on the rise like people yeah
people like that's why it rose that way yeah
exactly right so like
I think we have to understand
the industrial revolution
like you said took time and the question is
did we learn from those lessons
and if we did
hopefully we Learned better and the second question is
there's a lot of questions
going back to what you said earlier
there's a lot of questions we have to answer
that we have to think about when it comes to AI
and security and how we function as society like
are we okay with uh
AI functioning and pretending to be humans
are we okay with AI as counselors
are we okay with
digital ownership by a few people and entities
um when that's gonna be your interest into society
like what does that balance look like
and my concern right
with policymakers on all sides with this
is that most are
uh illiterate when it comes to any type of
basis of understanding technology
even at its fundamental level right
they can't usually if
if if you talk to them about algorithms
they really can't really even their eyes glaze over
but I believe this is one of the most central issues
of our time right from yeah
um not only how we become more competitive
but what's the function for society
so I agree with you like
so I guess I like to reframe the narrative to talk some
to bring up some things you said yeah
you said threats don't wait for permission right
you said that often right
what threats do you see with AI
and what steps can we take
maybe not as society cause maybe we're gonna be some
a ways away from and I think we are just
I'm just gonna just bite the bullet of reality right
we're some ways away from uh
America at least
having a direction
on what they wanna do in terms of policy
but what can individuals do
what can the parent do to protect
to make sure that their children are more uh
are not losing their critical thinking skills
or not going down rabbit holes
what can we do to protect our
our elderly from deep fakes
with fake voices that sound like
that just are just like me
but are leading people towards uh
you know uh
giving away their money or losing their security
like what advice
what practical advice
can we give people in the age of AI
right now yeah
so I think the most practical advice is
do you have to move this fast
like whatever moment that you're in
just pause a little bit right um so
when so there's an awareness issue
so what you're talking about
so there's two intersections of cyber and AI that
that keep me up at night right
just two yeah
just two
the big these are the big buckets right
the one bucket is
that this algorithmic power is resting on us
incredibly vulnerable to exploitation uh
system right
cyber space is
a space in which technology is easily exploited
we hear about it every day right um
so we are now gonna layer
we haven't solved basic cyber security
no we haven't
right now
we're going to layer
this very powerful decision making
uh capacity
it's not a tool we'll come back to that later
no it's not
it's an agent capacity yeah
it's an agent um
I'll tell people they're wrong when they say
and we're
we're gonna put it on a super vulnerable space
so let's just say that a particular set of agents
what whatever you ask them to do
whether it's an economic impact
a health impact etcetera
if they get really good at this
what we have found in human history
is the things that generate wealth
also generate contest competition
yeah over them
so if the algorithm itself is an agent that that um
uh is doing really good things
but it's sitting on a vulnerable origin and a
and a
and a superstructure that's vulnerable to exploitation
you're gonna see it so you
you know we uh
you saw just a week or so ago anthropic
yeah had a
you know a case with
with Claude and maybe we can talk about it a little bit
but so that's the one bucket yes
that this is technically vulnerable space
yes and now
we're going to put even more capacity on top of that
yes more responsibility that
that's worrisome to me yes
you know because we haven't solved the basic
cybersecurity problems now
this move to cyber persistence
which is the work that we've done here at university of
Cincinnati and that
I went to US Cybercom on loan from the university
and help develop this with
two colleagues from the Command
Emily Goldman and Mike Fishkeller
and that theory helped create a
a paradigm shift away from deterrence
a reactive uh
threat based uh
coercion model to hey
as you said uh
I think quoting me uh
you know that threats don't ask for permission
it's a continuous space of challenge
and therefore the only way
we secure the space is if you can anticipate
exploitation before it occurs yes
so it's a
it's about initiative persistence is what we argue
do you have the initiative
the second piece so that's where
that's where I worry about the
the security stuff
because we have not seized the initiative sufficiently
in and I want to come back to
and deep down go to your second place
and I want to talk about
so we've got that issue still hanging for us right
the second though is I think we have
we've been applying the wrong paradigm
so we need another paradigm shift with regard to
and you you touched on it
I'm in the camp that these
this is not a tool that you employ to
to use right
this is actually a decision making capacity
absolutely it is and if you understand decision making
as a loop of what we call the Udal loop right
you observe you orient
you decide which to prioritize
and then you act
Orient Orient
decide and act okay
and you do we do that as humans
like just in this loop continuously
and as soon as we act
we observe what came out of that action
we orient ourselves to that new observation
we prioritize what we should do
so right now your listeners are sitting back
and they've been observing us
they're orient now
they'll have to decide whether this is sufficient
for them to keep listening to us right right
hopefully you keep listening
yeah so I'll listen a little bit more
so then they reorient right
to listen more okay
so we're doing this unconsciously all the time
true
a these algorithms
have the capacity to do all of those functions right
so I can collect data observe yes
I can have the algorithm collect that data
and analyze it that's the Orient right
I can take it it can collect
analyze and prioritize
that's the decide and then I could have it collect
analyze
prioritize and act
what people call an autonomous vehicle
I prefer to call an AI chauffeur vehicle
right because it's an agent
it's a chauffeur it's just
it's a piece of code chauffeuring you
rather than a human correct
who doesn't like to get chauffeured in a car
we we we
we pay extra money for that
so
that's a model in which the algorithm has been given
the whole enchilada correct
it's gonna do everything correct
we can put it in very specific parameters to do that
but here's the key Robin
this is where I think the conversation's gonna go
and this gets back to your
all the way down to the parent level
what role and responsibility
do you want to give this algorithmic agent hmm
and that's where human agency comes back into this yeah
I can decide that I want it to
help collect information for my kid's homework
but I'm gonna I'm gonna set the parameter that
it's not writing it for you yes
so therefore you got to take
this information
saved your time in collecting the information
but now you're on your own right yep
I may be in a business where I say
collect this and analyze it for me
I'm gonna be the one that prioritizes
based on my human intervention
right right
thinking about the information that's in front of me
or I might say hey
prioritize this for me
but I'm gonna wind up being the act
I'm gonna be the one that makes that final action
to make the decision we
we have this human agency in our relationship with
the algorithm and so
my second big concern
is that we keep thinking about this as another tool
another piece of software
it's definitely not it's an
it's an agent that can it's an agent that's going to
it's but we get still hopefully right
we're in a situation where we get to decide
where its role and responsibility becomes
that's the hope if we adopt that framework
that's my paradigm shift for AI
then we get to accountability yes
so before you but you want to riff on anthropic
that's the big issue before you go there
before you go there I want to
cause
I want to talk about anthropic and bring up their side
but also kind of challenge part of what you said
yeah and yeah
just just
cause I believe just having just a full conversation
so let's let's
let's go through the anthropic um
which obviously was a security failure
um anthropic's uh
response is they were able to use AI to
to figure out that a what AI was doing
this is what they said right
yeah yeah yeah
I read the report yeah yeah
and then to their defense
they tend to be the most transparent with their AI
with their policy like I know what their policy is
they have a constitution they work to root it to
to root their agents in their
in their policy and
and just to explain to the listeners
you know and I'll just try to do it at a very
very low level um
this is why we tell people when you deal with models
the model you can't rely on the model for your security
right right
and so the model did have safeguards
but what happened is there were individual
kind of
agents that were doing seemingly harmless things
uh huh
but the orchestration
that just means the coordination of the agents
happened outside of that
and then they did the harmful things
I think the challenge
with some of what you said is that yes
there were there was harm there uh
uh
Chinese actors pretty it's pretty been
it's been pretty well substantiated that it was
a state sponsored attack
and we don't know all the big attacks that they made
but they were big attacks
why why it's so
I think concerning
is that 90% of the actual work for attacking these
targets was done by agents 90% yeah right
um which only 10% of the work was done by humans
so I think what this um
portends for the future
is that this is going to be democratized
and be easier for people to do
it won't even have to be state sponsored
it could be a small group of people
sure over time attacking systems
I believe the challenge is this
and I and I agree with what you're saying
I think the challenge is that
if we have to both become a AI digital
digitally literate that's what our parents have to do
we also have to use AI within our systems
because AI is attacking our systems
and yeah and so like
how do we balance my question is how do we balance that
because it's
I agree with you to a to a point
but it's like that kind of
that cat's been released out of the bag
probably 10 years ago so
how do we protect ourselves while still
remaining in control like
I feel like that's the tough balance
we haven't figured out yeah
so I this is where again
I think the reason why I'm proposing this
this responsibility
roles and responsibility model is it
it gives people the opportunity to think through
more in a terrestrial kind of context yeah
like if if
if you and I are right that this is about agency
we've been dealing with agency for
for all of human history it's called human agency yeah
and we've built rules
and regulations around those agents
just because it's code
we don't necessarily have to
to to to change that approach
so for for before we go to the anthropic
let me just stay with that car yeah
example sure
so we built a this licenser model that says well
over time we realized you gotta be of a certain age to
to drive a car
and you gotta go and do all of this testing
and it's it's pretty standard and
and then there's these standard tests you have to pass
right and then you get a license
yes to operate that vehicle yep
so a roles and responsibilities model would say
for AI chauffeurs
we need to license them yeah
and so
that would say that I would need to know how the AI
was educated stop talking about data
training on data and stuff like that
we're educating the agent yes right
so if you want me to trust
to get behind in the back seat of that algorithmic uh
uh chauffeur
why why is that a good chauffeur
well I can't go and test every single one as a
you as an individual user
so there's gonna be a whole new market
for licensing algorithms in actors
as actors yes
so on the medical side why
why should I trust a AI nurse practitioner
who's gonna license them
that's the way I think we need to
start to think about who's gonna license them
what's the what's the audit gonna be like
how do we how do we get there
and why do we do that it's
it's back to just basic trust trust
I mean right
so when I walk into my doctor's office
why do they still have their diplomas up on the wall
they have their diplomas on the wall
as a legitimizing function
to make you relax that
that somebody is credential to me right
that I went to this school and you know as you know
not everybody
looks at schools the same way we
we we suggest there's elite schools
there's research universities
like you see different types of teaching schools
so I affiliate with them because they legitimize me
right I think we can apply the same
model to algorithmic agencies
and that's a whole new market and that's a growth area
yeah right
to try to build build that trust
so when we get to Claude being used by a foreign
um intelligence agency
what is the roles and responsibilities model
illuminate for me right
well
if an anthropic employee had engaged in on
during their work day right
had engaged in the breaches of other people's systems
on behalf of a foreign uh uh
entity
they would be fired
they would be prosecuted under two laws
the Computer Fraud Act and the Foreign Espionage Act
right and they'd go to jail right
if I was a human agent doing that correct exact thing
so the question then becomes
how do you put Claude in jail
right because Claude now in this case
it wasn't total autonomy right
but it it portends
as you said where we're going yes right
so eventually so as a and
and right now what was the
I give anthropic a lot of cred for being open
right that these are breaches
etcetera etcetera
but it was also like the shrug emoji yeah
you know it's like yeah
well you know
it's like well
it happens it happens
it happens like yeah
somebody gets murdered every now and then things happen
yeah yeah
you know so these other systems got breached
what are you gonna do kind of like
what are you gonna do right
we this is a problem Rob
right because if we end with just the shrug emoji
if I'm sitting in another oh
let's just say Beijing and I decide well
that worked like
why would I use risk human agents
that's exactly if the truck spy in that way
let me keep using that stuff
that's exactly what that was gonna happen right 100%
so where's the accountability
now I'm not
let me just run with this as a little crazy thought
right but so Claude was not designed by anthropic
it was birth by anthropic
if we want to really go full on that
these things are agents yeah
right then these aren't developed tools
these are decision makers
yeah right
so there's they're creating absolutely decision makers
now is there a parental responsibility
well but
but before you get and we have all set of laws yeah
before that before you get to parental responsibilities
I I think the the
the real the biggest challenge is
and what you're getting to is
there needs to be universal policy for these things
because we have to start to get to it
because if we're not like
you know you know
anthropic would would say
we've been transparent with everything that we're doing
you know meta won't even release its policy yeah
and let's be fair right right
we're talking about them because they went public
because they went public right
and
and I'm not saying they don't deserve accountability
they do yeah
what I'm saying is like
all of these folks are accountable to some level of of
of having these uh
large algorithms that they
that they
we don't have any level of framework for accountability
yeah right
and what you're getting at is absolutely correct
and what people need to understand like yes
you
there needs to be some level of accountability for this
and so what and
and what would that be if we're talking big tech
um simple finds
the money's just yeah
it's a drop in a bucket most of the time
so let's just play this one out
and again I'm not recommending this
I'm just saying if you adopt a
roles and responsibilities model
as you look at this tech
all of a sudden
your brain starts to think in these terms like okay
well Claude went and did something illegal uh
a human would go to jail what
what's the equivalent right
putting Claude in jail well
it's probably saying that
that version of Claude should get off the market
yeah that sounds radical right
but that's a hey we're here for disruption yeah
so if you had that model now
you have a huge incentive for the company to go fix it
yes not not at the margins but go and fix it yep
and these these companies are great innovators
they have the capacity if you have
they always will adjust to market parameters
your your refrigerator
has more safety and security regulations on it
than your operating system on your computer oh
that's a line okay
last time I checked
people are still buying refrigerators
yes right because we regulated it to make it safer
didn't kill that industry no
it didn't kill the ability to sell
so this idea
that even some parameters about safety and security
that the tech has to
to fit into is somehow gonna kill off innovation
there's no sec economic sector in human history
that's been that fragile no
there's not and and I don't think tech is that fragile
it's not I
I actually have super faith in how clever we can be hmm
and if we say you know what
and this is go back to your mom and dad with their kid
and looking at it
yeah
I'd rather have these things that are a little safer
and a little more secure
could we could we do that these
we could absolutely do that if
if we had to
and that's where the interface between government
democratic government and uh
the market has always existed right
and so you know
right now this whole idea of just a bottom up approach
so what's happening we have had no central legislation
even on the broadest parameters of safety and security
in cyber uh
and or um
specific to AI right
I think two years ago was the last time I checked
there were 62 bills in Congress that had AI
in some aspect in the Bill right
not a single one of them
uh got passed
right
what's happened is that we have a whole bunch of states
that have been passing AI
based legislation some of which is actually pretty good
yeah um
emphasizing stuff like what you talked about earlier
about deepfakes like
do I own my personhood sure
what's my digital personhood
yeah what if you robo call using my voice
like do I have a
an ability to sue you for doing that
the states are doing this
the problem with that Rob
though is we're gonna have a patchwork
and it's gonna be all over the place
and industry so I think there's
a moment with the states being disruptors or agitators
I will on the policy side that the solution set is yeah
we probably shouldn't have the states filling this gap
we should have I think you use the word universal yes
right
big tech and little tech should prefer that
you'd rather follow one law
yes than 50 different laws
so could we come to a condominium right
this is a moment where we with AI right now
we can get out in front of this right
and have some very light touch
basic parameters of guardrails of hey
we're gonna we need you guys to prioritize safety
exactly and we need you to prioritize security
there's two different things
but they can do it every other industry has done it
my concern is that there's legislation to kill off
all type of legislation with no plan to fill that gap
I think that's intentional
but so I'm all I'm all in on that
it's not the right solution to have 50 answers
but we have to have we have to have an answer
we have to have an answer
and we have to have a framework
and agreement for what we want to be
as a society yeah
with this technology I wanna get to a few
cause I know we're wrapping on top yeah
a few just and quickly
kind of answer some of these in a few rapid fire
you're an educator mm hmm
how does education change now with the advent of AI
and what should never change
so what should never change
and this is gonna sound like a
a faculty member from from Arts and Sciences um
it's it's all about critical thinking yeah
and so
I think there's one thing that everybody can agree on
the next uh
five to 10 to 15 years are gonna be massively fluid
like we don't know where
where this tech is gonna ultimately take society
so we've got a lot of uncertainty yeah
it's not sufficient to say to a student
come and spend all this money and learn this degree
become specialized in something and then good luck
we're throwing you out into this uncertain world yes
so I think as educators
we need to spend more time discerning
through our own specializations
how do I position you to deal with uncertainty
yes and that is that to me is the most critical thing
people who can adjust somewhat on the fly
yes again
this goes back to my compressed time Assumption
it's fluid it's change
but it is compressed change
you're gonna have to be able to adapt
and move pretty quickly so
there are skill sets in the way we think
and the way we look at information
this goes all the way back to your
you know your your truth questions
how do we get people to reflect when they
when reflection requires time
in an environment that's compressing time
yes that's that's the nutshell
but I think as educators
we need to spend more time on that
yes then thinking about four year degree major pathways
and I know this is maybe heretical um
but we need we need more non traditional pathways
it may be um
that you come to the university and spend two years
yeah right
and there should be certification
legitimization reward for that
um we have associate degrees that do that
but I think we need to think out of the box
even more about that
but that you easily come back into the university
right exactly
you so
if you go out into the workplace for two years and say
you know I've gained that experience
now I wanna go and and get more education
I need you seamlessly coming back into our institution
and we're true
I don't need you to go through applications again
I don't need you to go through fees again to get in
we've got to make this seamless and fluid space
and I think with that there's also
you said it right about being a critical thinker I
I I often talk about artificial intelligence as uh
humans should think of it
think of it as artistic intelligence
meaning you are yeah
you are dancing with the algorithms
but you're not defaulting to them
you need to learn how to uh
you can use it to retrieve information
outline information uh
but using it as your source of decision making right uh
will not only uh
erode your brain uh
but it will also make it less effective like
because we are at the time where we have to now
it's not just about what I think does change
I would say is
there is no longer a need to simply
memorize a lot of information right
you have to actually learn the concepts and connect it
which is theoretical actually
it's harder right
yeah because
because memorizing things are easy
but we've been taught over years and years
because we had to do this as humans uh
we had information processors to memorize information
I think it is important to know some things
and memorize some things
but it's gonna be more important to know how to
when you get that information'cause
you can retrieve the information now
without any effort right
what do you make of this this information
so how do you connect multiple points together
and see something new
that's gonna be the challenge that you talk about yeah
no absolutely
and and so
you know we
when you look at a university
and it's not just UCLA around the world um
I had the privilege of
of working at Oxford over in England years ago
and you walk past buildings that were built in 1318
as the cornerstone right
that's a long time ago yes
Oxford is operating
pretty close to the way it operated in 1318 yeah
like we're still medieval
feudal and you know we've got our specializations
we call them colleges majors right
we're very segmented
we're we're in a world in which
interconnectedness is now the structural
organizing principle hmm
that's what cyber space is there it is our
the work that I've done is
that puts you in constant contact with everybody yeah
not episodic contact not over the horizon contact
right but constant contact alright
that's that we have to get people to UN
we have to shift our way of teaching right
and the product of teaching again
is more about adaptive thinking
how do we use these things
so I'm there's gonna be a whole bunch of human things
that we're out of the loop on
yes because the algorithm and sensors
and there's when we use the term algorithm right
that's it's a big term yeah
but you know
the real linkage is taking that compute power
tying it to sensors so I swallow a pill
and it's diagnostically able to give the doctor
a read on everything that's happening to me
entirely in my body we're
we're terrible at telling doctors what hurts us yes
we are when we go in
it becomes this translation problem
yes right
so a lot of health issues are
that first moment
where we can't explain what's going on
and the doctor has to like
go through hieroglyphics to figure it out yeah
so if I can swallow a
a sensor and have it in me for a couple days
and then it gives them entire diagnostic
I've solved that problem yeah
so take me out of the loop
cause I don't explain my health problems well enough
right but that doctor do we want him out of the loop
no no
so we have to figure out what on the loop looks like
yeah where that interface between the human
in this case doctor and this AI assistant yes
that has read more medical journal articles about my
inflict you know
my problem than the doctor ever will can observe cases
more cases than that doctor ever did
so in a way that algorithmic assistant is more expert
yes than the doctor
but we're still gonna trust that human
absolutely and so
we're gonna have to figure out how you leverage that
right
and the doctor doesn't lose their skills of thinking
so I'm gonna wrap on this question
that's the on loop that's the on loop yeah
and in in one sentence
how do we keep humans in the on loop
we have to assert ourselves hmm
right now we are way too passive
we are we are just letting the tech run
and we're not talking to our representatives
who don't understand it we don't understand it
that's okay
you can understand roles and responsibilities right
so let's start talking about it
in roles and responsibilities
do I want that algorithm to do this
hmm okay
why what does that make it better
I actually want a world where they're driving cars
cause we're really crappy at driving cars
we kill 47,000 Americans a year driving cars
what if we just cut that they don't need to be perfect
what if we cut that in half
that's 20,000 people who are still living yes
who
all of a sudden produce something in society that they
that we would have lost as a society
right cause they were killed in a car accident
so if I can cut car accidents by 50%
by connecting these sensors and these algorithms
yeah let's do that
yeah right
but let's figure out the public policy
absolutely that gets us there
I agree and you can apply
that to all these different functionalities agree
Richard Hartnett it was great having you on the show
definitely we could have had this go on a lot longer
yeah um
we appreciate all you've done for really the world
The University of Cincinnati uh
we're here of course at 18
19 and as always keep disrupting