00:00:00
thank you so much for joining us Mark
00:00:02
um our goal on the future proof series
00:00:05
is really to help create new knowledge
00:00:08
around the invisible forces of data
00:00:11
science and Ai and their tangible impact
00:00:14
on science society and the economy and
00:00:18
we're very excited to have you joining
00:00:20
us today to share your perspective on
00:00:23
how best to apply and really hone the
00:00:26
past decades incredible rise in computer
00:00:28
power and data and scientific
00:00:30
breakthroughs for the betterment of
00:00:33
society and so to start it would be
00:00:36
great if you can tell us about the goals
00:00:39
of the Alan Turing Institute where you
00:00:41
serve as Chief scientist
00:00:44
so the the Alan Turing Institute was
00:00:48
established by the the UK government
00:00:51
to make big leaps
00:00:55
um in what was then called data science
00:00:59
um to make the world a better place
00:01:03
um and of course if you think about Ai
00:01:08
and the big Renaissance that we're
00:01:11
experiencing Nai
00:01:13
um most of it is driven by the
00:01:16
availability of data and lots of it so
00:01:21
the the main goals of the island Turing
00:01:24
Institute are really to advance research
00:01:28
in data science and the EI and apply it
00:01:33
to some of the the biggest Global
00:01:37
challenges uh that we uh as societies
00:01:41
face
00:01:43
and so at the moment the Alan Turing
00:01:45
Institute is is about to announce its
00:01:49
strategy
00:01:51
um The Institute strategy and it will be
00:01:54
doing that in a couple of weeks at the
00:01:56
eiu key
00:01:58
um meeting and there we will be
00:02:02
announcing our grand challenges that we
00:02:06
will be focusing both data science and
00:02:09
AI on and those grand challenges will be
00:02:13
improvements in environment and
00:02:15
sustainability
00:02:16
improving the defense of the UK and its
00:02:20
security
00:02:21
and improving Health outcomes for the
00:02:25
population as a whole so those are the
00:02:27
goals of the Allen tuning Institute
00:02:30
a very very important goals to tackle
00:02:33
and and you know today we're closer than
00:02:37
ever to what Alan Turing imagined in his
00:02:41
very Landmark research paper from 1950
00:02:45
this world where uh we're interacting
00:02:48
with machines that think and in the wake
00:02:51
of all the Chachi PT and gpt3 news and
00:02:55
we can we can just type sort of a
00:02:58
message to a friend and and serve up a
00:03:03
Shakespearean play a research paper
00:03:04
lines of code
00:03:07
um it'd be great to learn about you know
00:03:09
some of the the ways you think these
00:03:11
capabilities will inspire creativity for
00:03:15
their scientific discovery or lead to
00:03:18
other beneficial outcomes
00:03:20
so I think the first thing is to see
00:03:22
that um with the machines that we we
00:03:25
work with don't think
00:03:28
uh and I think we need to really be
00:03:31
quite grounded in that that we don't
00:03:33
have
00:03:35
um intelligent machines we have very
00:03:38
powerful computers that still follows
00:03:41
Moore's law in terms of the number of
00:03:43
transistors that actually uh go on to uh
00:03:46
the Silicon but the number of
00:03:48
transistors that can go on to Silicon
00:03:50
are enabling the compute capability
00:03:55
um to to provide us with power uh to
00:03:59
process data at scales that we've never
00:04:01
been able to think of before and so for
00:04:04
example things like cat GPT and gpt3
00:04:10
um are the outcomes of just very very
00:04:14
large models of language that have been
00:04:19
um whose parameters have been estimated
00:04:21
uh by you know pretty much crawling the
00:04:25
whole of the globe's internet
00:04:29
um and so so that's hugely exciting that
00:04:32
that we have that capability to be able
00:04:35
to do that
00:04:37
um which 30 years ago we couldn't
00:04:41
and so I think that in terms of the
00:04:44
important ways that these capabilities
00:04:48
um are going to inspire creativity
00:04:52
enhance the scientific discovery process
00:04:55
and and lead to beneficial societal
00:04:58
outcomes I think we need to look at
00:05:02
um
00:05:03
you know some of the advances in
00:05:06
technology that we've seen previously
00:05:11
um and if we go back if we go back 30
00:05:15
40 years
00:05:17
um to the Advent of personal computing
00:05:23
a computer and the capability to program
00:05:26
that was being put on to individuals
00:05:30
desks onto their
00:05:33
um you know their kitchen tables
00:05:36
and that then just you just you saw an
00:05:39
absolute
00:05:40
exclusion
00:05:42
creativity people writing programs to
00:05:46
um to do home accounts people writing
00:05:49
programs to manage lists to you know to
00:05:52
to communicate and so on and to write
00:05:55
programs to to play games and so on
00:06:00
um and there were huge numbers of of
00:06:03
advances that that capabilities that we
00:06:07
you know we just didn't really expect
00:06:10
and then if you fast forward to the next
00:06:12
big
00:06:13
um
00:06:14
technological advance
00:06:16
the the internet
00:06:19
and again you think of that Global
00:06:22
connectivity that we then have
00:06:26
um the ability to access information
00:06:29
that we had to physically go to
00:06:31
libraries to get
00:06:33
um I know that is online it's streaming
00:06:37
um and and again it just opened and
00:06:41
tapped into you know the the the the
00:06:44
creativity of of whole populations
00:06:50
and then if you think of
00:06:52
the iPhone or mobile mobile Computing
00:06:57
and taking that capability offer
00:07:00
desktops and basically putting it into
00:07:02
our back pockets
00:07:04
and that Mobility uh again
00:07:08
took us to another level of of
00:07:11
creativity and and you know the various
00:07:14
applications that that we have
00:07:18
um and so
00:07:19
so I think that
00:07:22
um these advances these recent advances
00:07:25
in AI
00:07:27
um are going to
00:07:29
um give us uh the the tools you know to
00:07:33
harness Mankind's creativity
00:07:37
um in similar ways to personal Computing
00:07:40
or the internet or the iPhone and it did
00:07:43
previously and that means in ways which
00:07:46
we probably can't really you know think
00:07:49
of ourselves at the moment
00:07:52
um but so so I think in terms of of
00:07:55
creativity we should really just be you
00:07:58
know just just watching and waiting and
00:08:00
seeing what's what's going to come out
00:08:02
of it and in terms of scientific
00:08:04
discovery I mean we've we've got you
00:08:07
know something like Alpha fold which uh
00:08:10
again has used the the compute power and
00:08:15
the power of data and combining those uh
00:08:18
to give us really smart algorithms or EI
00:08:22
algorithms to you know start to predict
00:08:25
the folds of proteins and so on and that
00:08:29
those tools are then being put into the
00:08:32
the hands of basic scientists who are
00:08:35
trying to understand the Genesis of some
00:08:38
of the you you know some of the diseases
00:08:41
that we face as as whole populations and
00:08:44
as mankind as a whole
00:08:47
um and I you know I I I think again
00:08:50
we're going to see a supercharging of
00:08:53
some of the advances in scientific
00:08:55
discovery because
00:08:57
of the of these let's call them AI
00:09:00
capabilities and then again in terms of
00:09:04
beneficial societal outcomes I I think
00:09:07
in terms of
00:09:10
um you know either if we think of some
00:09:12
of the
00:09:13
Grand challenges uh in in the island
00:09:17
Turing Institute you know defending our
00:09:20
our nation
00:09:22
um making you know the population secure
00:09:25
improving Health outcomes improving our
00:09:28
environment mitigating you know our
00:09:31
sustainable uh infrastructure against
00:09:35
um climate change and so on I I think
00:09:38
that that some of these AI tools are
00:09:42
going to lead uh to beneficial outcomes
00:09:45
uh in some of these uh challenge areas
00:09:49
hmm thank you that's that's very helpful
00:09:52
context and perspective and and you're
00:09:55
absolutely right that we don't quite
00:09:58
machines that think and but we do have
00:10:00
machines that appear to be thinking and
00:10:03
um really sort of outpacing public
00:10:06
understanding and it would be great to
00:10:08
to get your take on you know some of the
00:10:11
risks that you think are important to
00:10:13
bear in mind for different audiences be
00:10:16
it businesses or end users or
00:10:18
entrepreneurs
00:10:20
so one of the the major risks that we
00:10:25
face at the moment
00:10:27
is that
00:10:29
um I mean by and large
00:10:32
the scientists like myself who are
00:10:35
developing these AI algorithms
00:10:38
um are not entirely clear as to how they
00:10:41
work
00:10:43
and therefore how they might feel and
00:10:49
um I think that the the big risks one of
00:10:53
the big risks that we're going to face
00:10:54
is that we are putting these potential
00:10:57
tools
00:10:59
um into the hands of policy makers you
00:11:04
know governments
00:11:06
um and it's not entirely clear what
00:11:10
their failure boards will be and we've
00:11:12
seen some very high profile examples of
00:11:16
of some of the failures of of these big
00:11:19
large language models
00:11:22
um the the potential for bias because
00:11:26
they're being trained on well we're not
00:11:28
entirely sure because it's just a you
00:11:31
know just a huge scrape of of the global
00:11:35
internet
00:11:37
um so I I think that
00:11:40
um the introduction of these sorts of
00:11:42
biases and how we understand what those
00:11:46
biases are how we engineer them out
00:11:50
um is is something that that is going to
00:11:52
be a big risk
00:11:54
um the uh I I think the the failure
00:11:58
modes uh how sensitive these
00:12:02
um AI methods or algorithms or tools are
00:12:05
to small perturbations in operating
00:12:08
conditions and they can move from
00:12:10
something that is operating well you
00:12:13
know giving good results giving you know
00:12:16
sensible answers to something that that
00:12:19
you know could could be wildly you know
00:12:23
weird uh and of course potentially
00:12:27
dangerous
00:12:28
so I I think that we we really need to
00:12:31
be working
00:12:32
um you know very very closely with with
00:12:35
our government and understand on the one
00:12:39
hand what are the potentials what are
00:12:41
the opportunities but on the other hand
00:12:44
what are the risks uh and what are the
00:12:47
dangers of of adopting this and and
00:12:51
if you think of the history of flight
00:12:55
now when we started building aircraft
00:12:57
didn't really understand
00:13:01
um all of the technical details of you
00:13:03
know why planes stayed up in the ear
00:13:07
um and by and large we still don't know
00:13:10
uh fully but it it doesn't stop us uh
00:13:15
safely using aircraft and using flight
00:13:18
for transportation
00:13:20
but
00:13:21
to get to this level of safe usage of
00:13:26
aircraft
00:13:27
we've had to learn some very very
00:13:31
serious lessons because of some of the
00:13:34
disasters uh the aircraft and Airline
00:13:37
disasters that that um have you know
00:13:40
we've experienced over
00:13:42
um well the whole history of of flight
00:13:45
and so what we ideally would not like to
00:13:49
do is be in a similar situation where we
00:13:52
have to rely on disasters happening
00:13:56
um to better understand and then better
00:13:59
use some of these AI tools
00:14:03
so so I think entrepreneurs
00:14:07
um businesses and end users are all at
00:14:11
the same levels of I suppose
00:14:13
entrepreneurs and businesses are an even
00:14:16
greater risk because their their
00:14:17
businesses are going to be based on
00:14:19
something which you know has inherent
00:14:21
risk as well potentially as inherent
00:14:23
risk associated with it and so I think
00:14:26
the mitigation of those risks at the
00:14:29
design stage
00:14:31
um and then the the deployment stages is
00:14:33
going to be really critical
00:14:36
um yeah and and building on this
00:14:39
thinking around opportunities and and
00:14:41
risks what do you think are some of the
00:14:43
most important areas for research and
00:14:46
this this significant inflection point
00:14:49
I I mean I I think the the whole notion
00:14:53
of the robustness
00:14:55
um you know how how robust are are these
00:14:58
tools going to be
00:15:00
um
00:15:02
uh the the you know will always operate
00:15:05
in a safe way and we don't you know we
00:15:07
don't have anything that that flips from
00:15:09
uh you know something that's safe to
00:15:12
something that is unsafe so there's an
00:15:15
awful lot of research going on at the
00:15:17
moment looking at robustness
00:15:21
um of some of the the Deep architectures
00:15:24
uh associated with things like um GPT
00:15:29
gptc and so on
00:15:31
I think another really important area I
00:15:35
mentioned bias and really understanding
00:15:38
the importance of data and and of course
00:15:43
the input of humans into that that we
00:15:47
we're not taking human I mean that's why
00:15:50
track gbt is so good is is because they
00:15:55
have used humans to to actually you know
00:15:59
help out in the tuning and the learning
00:16:02
and of of the uh and the training of
00:16:05
these tools so having humans actually in
00:16:08
that loop at some point
00:16:11
um and being able to
00:16:13
um you know to use that is is going to
00:16:17
be really very important I would say
00:16:22
um thinking about research and perhaps
00:16:25
learning more broadly and how do you
00:16:29
think pedagogy will be impacted by these
00:16:32
inhibit movements so again you know if
00:16:36
if we go back to
00:16:39
um when calculators first came along
00:16:43
[Music]
00:16:44
um
00:16:45
you know
00:16:47
what was the net effect of that well you
00:16:51
know mental arithmetic is not so
00:16:52
important anymore
00:16:55
um but what is important
00:16:57
um and what was more prominent was that
00:17:01
the education in terms of you know the
00:17:04
the the basic mathematics
00:17:07
um could progress
00:17:09
um without having to focus on the root
00:17:12
learning of of you know mental
00:17:14
arithmetic why because we have these
00:17:16
tools these calculators that do that
00:17:18
mental arithmetic photos and so I think
00:17:21
that we'll see similar
00:17:24
advances
00:17:25
um in in terms of education and and
00:17:28
pedagogy
00:17:30
um uh I mean what one example one of my
00:17:33
former PhD students was he was now a
00:17:37
professor at a university he was writing
00:17:41
a a research Grant application
00:17:45
and he used chat GPT to to just to write
00:17:50
all the sort of boilerplate text that
00:17:52
that's required
00:17:55
um and he could then focus on you know
00:17:57
making the the more nuanced arguments
00:17:59
about you know why this research is
00:18:02
important and why it should be funded
00:18:05
um so I I think like like most advances
00:18:07
in in in in technology once we get these
00:18:12
tools
00:18:13
um then we need to we need to revise uh
00:18:17
in terms of what is important uh as far
00:18:20
as pedagogy and as far as the necessary
00:18:23
skills are concerned and then develop
00:18:26
those
00:18:27
so so I I think you know we run for an
00:18:30
exciting time
00:18:32
um as far as yeah as far as learning is
00:18:35
concerned I I'm not concerned about it
00:18:38
you know I think it's I think it's a
00:18:40
good thing
00:18:41
um and and it will be very exciting I
00:18:45
think that will be comforting for for
00:18:46
many to hear
00:18:48
um and I suppose you know taking a step
00:18:51
back I I understand that your background
00:18:54
um as a civil engineer and both at IBM
00:18:56
and as a professor of civil engineering
00:18:58
at the University of Cambridge has
00:19:00
influenced
00:19:02
um your interests in mathematics
00:19:05
statistics and Engineering uh more
00:19:08
broadly and that you sometimes describe
00:19:11
this intersection as data Centric
00:19:14
engineering and I'd love to get your
00:19:16
take on what that means and and why this
00:19:20
is a really important area
00:19:22
so um so first of all I should I mean I
00:19:25
should just confess to say that I'm not
00:19:27
a civil engineer
00:19:29
um so it's quite it's I'm usually
00:19:33
introduced by my colleagues at
00:19:34
cambridge's
00:19:36
um this is Mark he's the professor of
00:19:38
civil engineering that isn't a civil
00:19:40
engineer
00:19:42
um so I I actually was was brought in to
00:19:45
Cambridge from the mathematics
00:19:48
department at Imperial College where I
00:19:50
was professor of statistics
00:19:54
um but I I think that um and and so it's
00:19:58
a good example of this intersection of
00:20:00
mathematics statistics and engineering
00:20:02
and I think it builds on
00:20:05
what is happening in in EI
00:20:09
um because you know most of these AI
00:20:12
systems like chat gbt
00:20:16
um are in essence big statistical models
00:20:20
right so they are at the end of the day
00:20:23
you know learning the regularities of
00:20:25
some sort of language
00:20:29
um they are using mathematical methods
00:20:31
to
00:20:32
learn their parameters
00:20:35
and then there's a huge amount of very
00:20:38
clever Engineering in in developing
00:20:41
these large-scale systems
00:20:45
so so I think the the intersection of
00:20:48
mathematics statistics and Engineering
00:20:50
computer science it's a hugely fertile
00:20:54
area and
00:20:56
um I talk about data Centric engineering
00:21:00
uh in two ways
00:21:02
so the first one is is the data Centric
00:21:04
engineering is losing you
00:21:07
so the Victorian Engineers were doing
00:21:11
data Centric engineering they were
00:21:13
conducting experiments they were making
00:21:16
measurements they were gathering data
00:21:19
and then they were defining empirical
00:21:21
laws of of whatever you know how
00:21:24
structures stand up how fluids for what
00:21:28
have you
00:21:29
foreign but what is no different
00:21:32
in a similar fashion to the the
00:21:35
revolution that we've been seeing in AI
00:21:37
is the availability of the amount of
00:21:39
data
00:21:40
and that we can have the amount of data
00:21:44
that is of our very very fine
00:21:46
granularity that gives us insights to
00:21:49
things as small as the sale right we can
00:21:52
gather data about the details of the
00:21:55
cell we can gather data about you know
00:21:57
the darkness of the cosmos and pretty
00:22:00
much everything in between and so that
00:22:04
data right uh has information about the
00:22:08
things that we want to study and and so
00:22:10
we're using that and putting that data
00:22:12
right at the center of whatever it is we
00:22:15
do whether it's Sciences like Alpha
00:22:18
holds you know they took all of this
00:22:20
data smart engineering of you know the
00:22:23
alpha fold algorithm put it right at the
00:22:26
center of everything that's being done
00:22:28
and it's the same with with engineering
00:22:31
engineering
00:22:33
um whether it's Aeronautical Engineering
00:22:35
whether it's agricultural engineering is
00:22:38
being completely transformed
00:22:41
because of this availability of data and
00:22:45
and you could even go further forward
00:22:47
and say you know it AI which is enabled
00:22:51
by data it is Data Centric is enabling
00:22:54
some of the you know the biggest
00:22:56
advances that we're seeing in
00:22:58
engineering science and then and then in
00:23:01
engineering practice
00:23:03
so it is very important
00:23:07
um excellent
00:23:09
um really fascinating
00:23:11
um I I'd love to hear a bit about how um
00:23:15
you know some of the projects that
00:23:18
you've worked on in the past
00:23:20
um that put kind of data at the center
00:23:22
of problem solving
00:23:25
um and are applied to some of the big
00:23:27
challenges uh that we were speaking
00:23:30
about earlier in this call yeah I mean
00:23:32
one um so so one area of AI that we
00:23:37
maybe haven't mentioned is this whole
00:23:39
idea of the digital twin or the digital
00:23:42
Avatar
00:23:45
and
00:23:47
um and what what does that mean well it
00:23:49
means that we have something in
00:23:51
in the world
00:23:53
whether it's an aircraft whether it's a
00:23:56
bridge
00:23:57
whether it's a process plant whether
00:23:59
it's a city
00:24:01
whether it's a person
00:24:03
and what we can do is we can develop and
00:24:08
realize
00:24:10
a digital representation of that entity
00:24:13
in the physical world
00:24:16
and we can couple them together we can
00:24:18
twin them
00:24:20
with data
00:24:22
so we can make measurements across our
00:24:25
city we can make measurements
00:24:27
of our aircraft we can make measurements
00:24:30
of our farm
00:24:32
and we can feed that data into the
00:24:35
digital representation
00:24:37
and then use the coupling between the
00:24:40
physical and the digital to better
00:24:43
control whatever it is we are interested
00:24:46
in or better design whatever we're
00:24:48
interested in if I give you one example
00:24:51
I'll give you two examples and one is in
00:24:55
agriculture so we you know we we have a
00:24:58
um
00:24:59
we have
00:25:00
real big challenges
00:25:03
um in terms of
00:25:05
future agriculture
00:25:07
and one area that is being developed is
00:25:11
the use of Hydroponics
00:25:14
um
00:25:16
and
00:25:18
um
00:25:19
there is a company
00:25:22
that has been developing
00:25:25
um
00:25:26
farming Underground
00:25:29
so not above ground but actually
00:25:31
underground and a Clapham Junction in
00:25:35
London
00:25:36
there's a disused true blind
00:25:40
which has a five minute
00:25:42
incredible
00:25:44
um it grows herbs it grows vegetables
00:25:48
that that Supply London
00:25:52
now
00:25:54
um
00:25:56
we were brought in to help with this
00:25:59
because growing
00:26:02
plants agriculture Underground
00:26:06
is completely different from
00:26:09
overground the way that
00:26:12
you control heat the way that you
00:26:15
control
00:26:16
um you know various gases oxygen and so
00:26:19
on it's completely different
00:26:23
um and
00:26:24
what what we were asked to do uh was two
00:26:28
things there was one
00:26:29
generate data from the farm uh about all
00:26:34
of these key indicators that were
00:26:36
important and being able to control how
00:26:39
the yield of the the farm would work and
00:26:43
then develop a digital twin of the Forum
00:26:45
so that they could better control uh the
00:26:48
operating conduct or the growing
00:26:50
conditions of the firing so that's
00:26:53
exactly what we did we built a digital
00:26:55
twin of the environment of the Forum
00:26:59
um of the way in which the the various
00:27:01
crops would grow
00:27:03
how they would use
00:27:06
um CO2 and so on and humidity and how
00:27:10
the farm would be able to reject the
00:27:12
retain heat and so on
00:27:15
um and
00:27:17
the use of that right with all of the
00:27:20
sensors feeding data into the digital
00:27:22
twin and the digital twin then saying
00:27:24
here are probably the conditions that
00:27:26
would be optimal for this this type of
00:27:30
yield
00:27:32
um we um we were able to and increase
00:27:35
yield by many percentage points uh and
00:27:39
and make that fund really efficient and
00:27:41
so so this is incredibly exciting
00:27:44
um it's now used
00:27:47
um you know the next time you go into a
00:27:49
michelin-style restaurant in London
00:27:50
you'll probably be you know eating herbs
00:27:53
that were grown uh under Clapham
00:27:56
Junction
00:27:58
um
00:27:59
and and that that um whole area of
00:28:01
Agriculture uh we're now working with
00:28:03
some of the the government uh
00:28:06
agricultural research stations where
00:28:08
they want to develop digital twins of
00:28:12
you know some of the the big
00:28:13
agricultural Farms
00:28:16
so that's agriculture
00:28:18
um another area uh is is our
00:28:21
infrastructure you know what uh the
00:28:24
roads that allow us to you know our
00:28:27
transportation system online to to to
00:28:30
work well the bridges that that Kari uh
00:28:33
are trained across you know rivers or
00:28:37
gorges and so on and
00:28:40
one one project that we've been involved
00:28:43
with uh to deal with Transportation was
00:28:48
uh when Network rail were building some
00:28:52
new uh real Bridges I can start a show
00:28:56
and
00:28:57
[Music]
00:28:58
um
00:28:59
what they did
00:29:01
they embedded a number about 180
00:29:06
sensors so these are fiber optic sensors
00:29:10
um in the concrete
00:29:12
of the bridge when it was it was
00:29:14
actually being constructed
00:29:17
and the bridges uh when they were
00:29:21
um
00:29:22
constructed when they were actually
00:29:25
deployed
00:29:27
they were described as living structures
00:29:30
because
00:29:33
we were able to gather this data from
00:29:36
the bridge
00:29:37
in real time and continuously so every
00:29:40
time a train went over it we were
00:29:42
getting all of this data and so what it
00:29:45
made what it it means is that we could
00:29:49
continuously monitor the performance of
00:29:52
that structure
00:29:53
without sending Engineers out to look at
00:29:56
it because we're getting this footprint
00:29:58
we're getting its heartbeat
00:29:59
on a regular basis
00:30:02
and again what we did is we built a
00:30:04
digital twin of that structure and the
00:30:07
continuous speed of data into the
00:30:10
digital twin and the digital twin then
00:30:13
being able to see or answer questions
00:30:17
about
00:30:18
is the bridge performing well as the
00:30:21
bridge degrading is there going to be a
00:30:23
point where we might start to see
00:30:26
structural
00:30:27
um problems that could put potential
00:30:30
users at harem
00:30:32
and so we have this now where the the
00:30:36
bridges remove is is is monitored
00:30:39
continuously
00:30:41
um it is controlled continuously and it
00:30:45
is all done remotely
00:30:48
um so so this notion of a digital twin
00:30:51
you know feeding off the data uh that
00:30:54
comes from that bridge and so this is a
00:30:56
complete transformation in the way that
00:30:58
that Bridges and critical infrastructure
00:31:01
can be operated and can be controlled uh
00:31:06
to both
00:31:07
make them more efficient
00:31:10
in in ensuring that their availability
00:31:12
is is is optimized
00:31:15
and makes them look safer right so that
00:31:18
if there are any
00:31:20
um
00:31:20
faults that are starting to put the
00:31:24
structure onto a pathway where me well
00:31:28
um feel we can see that long before that
00:31:34
critical point happens and and I'm sure
00:31:36
you've seen
00:31:37
very recent stories of some of the the
00:31:41
disasters uh the the the bread
00:31:44
genetically that just completely
00:31:45
collapsed
00:31:47
um and and you know the mortality was
00:31:50
was was was Dreadful having something
00:31:52
like this
00:31:54
um as I said will will make operation
00:31:57
more efficient
00:31:59
but also make it much more efficient and
00:32:02
much more safe and ensure the uh the the
00:32:07
the risk of of these catastrophic
00:32:10
failures uh and the devastation that
00:32:13
they cause uh is is is is is greatly
00:32:16
reduced
00:32:19
thank you thank you Mark um there's a
00:32:21
really fascinating examples of data
00:32:24
science in practice to to solve real
00:32:26
world issues
00:32:29
um to close it would be great to just
00:32:31
get your advice
00:32:33
at someone who wants to be future
00:32:36
approved as we call it and you know what
00:32:39
do you think are the most exciting and
00:32:41
important forces to stay knowledgeable
00:32:45
about and to continue to track I mean I
00:32:48
I think that you you you you you just
00:32:51
don't stop learning
00:32:53
don't stop reading and read you know
00:32:57
read widely read about politics read
00:33:00
about read about politicians read about
00:33:03
economics you read about Finance read
00:33:06
about engineering read about physics
00:33:08
read about anything
00:33:11
um and and just always stay informed
00:33:15
um I I don't think there's any one thing
00:33:17
I think it's just a case of
00:33:19
you know just being naturally interested
00:33:22
in the world around you and don't stop
00:33:24
asking questions I think if you do that
00:33:26
you're going to be future proof for sure
00:33:29
excellent well thank you so much for
00:33:31
your time and insight really appreciate
00:33:33
it
00:33:35
my pleasure
