Exploring how artificial intelligence improves efficiency while human judgment, creativity, and decision-making remain essential.
Join this webinar to understand the balance between AI and human expertise.
Presented by Karl A L Smith
In association with IABAC
Uh good afternoon and uh good evening to everyone uh from wherever you’re joining
us. Uh and uh welcome to our webinar series uh experts speak uh in
association with IAB the international association for business analytics certification and uh
let us start today by giving you a brief introduction about the speaker for today Mr. KL Smith. Uh KHL is the founder of
social media platforms like Mi Social and AllMe. And uh he has over 25 years
of experience helping global organizations uh transform the way they design, build, and deliver services
while also partnering with banks, governments, universities, and Fortune 500 companies to turn complexity into
clarity. And uh today he’ll be covering the topic on AI effectiveness and human
eminence. And uh just a request to all the participants uh please post your questions in the Q&A section and uh
we’ll present it to the speaker at the end of the presentation. Uh over to you K.
Hi uh everyone. Uh thank you for the introduction and thank you for the opportunity to share uh my musings or my
experience. Um I do have a short presentation. I don’t want to focus too heavily on the
presentation because actually it’s the ideas behind the images that are important rather than the images
themselves. Um my background in AI goes back to
around 2001. Uh and and realistically if you know what AI is um it probably goes back to
the first time I ever coded anything which was back in the 1980s. uh looking at how we can augment human
existence and that’s what I see computers doing rather than taking over from them. However, like everyone else
on the planet, I am aware that there are some issues with how we’re utilizing technologies and specifically AI. Uh
which is why I put this image together. So hopefully no one’s having a heart attack about now. Um but there is this
sense that AI is crushing humanity um up against the business need for
profit and actually it doesn’t have to um there does need to be a rebalancing I
think in how people operate how they expect to work and also what work should
be. Um there’s a great deal of work that’s been carried over from the last
century that frankly has never needed to be done by humans and should always have
been done by machines. Just like we realized during co we didn’t need to go
to shops but we could buy everything online. Um although someone like me had been banging on about using e-commerce
for 20 years, no one actually listened until there was a situation where they couldn’t go to a shop. uh same way AI is
uh dramatically changing a lot of entry-level work. Um and so instead of
fighting that, we need to make sure that we give young people a different entry point than doing manual tasks and give
them over to the skill sets that they really have as humans, which is innovation and invention. Um, I just
want to bear in mind what artificial intelligence actually is cuz there’s a
lot of misleading information out there at the moment. Uh, just because everyone
wants investors and I get that that’s um a human drive to say what we’re now
doing is AI. Um, but actually there’s a lot of systems that are currently being
purported to be AI which are not. So let’s start at the beginning. Artificial intelligence is a field of creating
systems that can perceive, reason, learn and act with a degree of autonomy. They
themselves can do that even at a finer level of the programming. They can
respond uh through data matching or through pattern analysis and change what
their actions are. Um if the system you’re working on doesn’t do that, it’s
not AI. Um, it may well mimic AI and there’s a lot of that that’s happened
and a lot of unfortunate people have been taken to the cleaners on investments for things that are not AI.
Um, go back, always go back to what does it do and what problem does it solve?
And this is where I come into AI in my experience. I’m looking to solve actual
problems, not find an AI and then look for a use for it, which is what I think
a lot of the scrabbling around at the moment is focused on. Um, I do want to
go back to uh, you know, where AI comes from. And AI is not new. It’s very, very
old. It’s older than I am. And I feel quite old. Um, for the Star Wars fans,
you’ll know what the first sentence means. Uh for everyone else, the key thing is AI is 75 years old or well 76
if you want to conclude this year is over but at least 75 years old. It’s
very very old as a thing. What has changed is access. And what we need to
be aware of as engineers and as architects and as professionals working
uh to solve problems is that there is now this sense that everything can be
done by AI and it can’t just like in the past everything can’t be done by
software. I really don’t want to get my hair cut by a piece of software. I really don’t want to be driven by a
piece of software and I can go into that later if people are interested. um uh given the recent announcements by
Tesla where they’re dumping their self-driving cars when in fact other people have solved that problem because
they looked at it a different way. Um so what are the AIs I’m talking about? So
these are the the the main AIS and there’s a lot of them and I probably missed some out and for the engineers on
the call. I did look at who was attending there does seem to be quite a few engineers. I can’t do what you do.
Uh I I am not the at the coldface uh solutions person. I’m at the
enterprise level. How does this interface with the vision of the organization? How does this fit the
outcomes we’re trying to achieve? How can we deliver the data systems necessary to underwrite these AIs? How
will we conform to regulatory compliance and governance? That’s where I sit in in
this model of delivering AI. Um, so I won’t be going into the actual
functionality at a at a macro at a micro level of how these things actually work
because actually there’s people on this call that probably know better than me. Um, never set yourself up for a fall.
Um, so I’ve I’ve called out a number of systems here uh as solutions. Um uh the
first one is expert systems, machine learning, probabilistic and statistical
AI. I always think of that as basin networks. I can’t stop myself because I’ve used basic networks. Uh
evolutionary and bio inpired AI, fuzzy systems or again um I tend to f focus on
the logic models, hybrid AI where you’re intermixing a number of AIs, embodied
and robotics AI, deep learning which I’d love to get into. It fascinates me, but I’ve never had a client willing to pay
for me to do a uh to solve a problem with it. Uh natural language processing
and large language models. Now, uh I’ve starred the ones I’ve actually worked on
or utilized. Um so, if I can move this out of the way of my notes. Um
my uh expert systems uh solution related to utilities management. And this was
back in 2005. As I said, AI is very old. Um, and uh the the thing that we were
trying to understand was how we could codify uh processes around utilities
management and utilities troubleshooting. Uh and what we did was we took a whole series of data around
the problems and process uh solving that was done to fix uh water supply issues
uh electrical supply issues and we codified them uh and then we made them
available to an expert system that could be responsive to new queries around
problems and therefore offer solutions to engineers. Now in some ways it does dumb down
engineers because they don’t then need to know everything. Uh this is field engineers but at the same time it also
lifts a lot of people away from hidden knowledge uh and really does aid people.
The machine learning uh piece I worked on was actually to do with uh anti-money laundering in financial services uh and
actually was looking at pattern analysis. Um although later I did actually do some training around IBM
Watson which is an amazing uh I uh IIA system. Um pattern analysis has been
around for hundreds of years. We’ve always used it. Um and in this instance
we were dabbling with how can we track unusual trading actions uh within
complex systems where you’re dealing with um essentially financial instruments. Um, and we utilized uh an
AI uh logic model for machine learning to resolve uh where these transactions
were coming from and also to look at trading platform activity to ensure that
uh insider trading and um uh certainly offsetting uh risk was managed
effectively or hedging. Uh so that was around 2006 2008.
uh probabilistic and statistical AI uh was something to do with basin networks I was involved with in 2015. Uh it was
with a company based in Australia who were looking to commercialize what had
been until that point a decision engine around uh the five eyes program which is
a military intelligence system. And what this was looking at was how uh the
system could draw data from other sources other than military systems to
perhaps predict the likelihood of uh someone making a claim on insurance or
how they might uh be considered to be a good risk for a mortgage and and
actually just looking at other ways. We did actually look at it also as a way to predict the next US president. uh and
and we didn’t actually get that right as um as Trump. So, it’s a very interesting
uh way of utilizing an AI. Um I do have diagrams of how that works if people are
interested. Um fuzzy systems that was actually for a very well-known UK
um um what’s it called? Uh property uh engine. And what they were looking at
was the problem they had was that a lot of data was available on uh properties
in the UK that were for sale that was conflicting and it was very hard to work out if they were all the same property
or not in a given area. And actually that was because the state agents were specifically hiding certain data points.
Uh and what we did was we utilized the fuzzy logic system to uh uniquely
identify properties and then make them available for the estate agents to then um be on that platform where they could
actually say they can advertise it. Very interesting idea. That was 2008. Again,
this is quite a while ago. Uh hybrid systems. Uh I worked with the expert
systems and machine learning for a military fire control system uh back in
2018. Uh and that was in uh Southeast Asia for a client that actually wanted
to support the observation of um military activity utilizing drones which
is all all the uh the main talking point these days in a lot of AI. Um but that
was back in 2018. We should also be aware that fire control has been uh
automated um for a long time with things like sea whiz uh in battleships. Um but there’s
always been a human intervention point with that uh control at least until you
get within a certain distance of a battleship. um an actual language processing uh that
was actually based on a uh case study I saw by someone else. I saw a case study
in 2006 of the Verizon uh natural language language processing system that
they were using to respond to support requests. And off the back of that, I
use that for a a European telco company as a solution to their problem of
dealing with millions of requests for support. And the last last one was
actually quite recent. So that rise from 2006, the LMM um thing, which is what
everyone’s really going on about these days. I was asked to investigate them as
a viable and um traceable way to do uh
biomedical science uh documentation uh in terms of clinical reporting and
that was 2024 again looking at um how we should
structure the systems in order to make sure that the LML llm is not fabricated
ating information but is uh providing a filter to uh focus the deliverables. And
that’s where I came across the notion of a narrow language model which would actually be used as a treasury index if
you’re going to use the the thoughts around uh information management uh or
as a core index in in creating reports which is where I think a lot of the
large consultancies have gone wrong. So they haven’t really investigated the technology uh in such a way to
understand how they can best deliver value uh to clients.
Now, um, so that was rather long-winded. I’ll try and speed up if possible. Um,
because I want to open up to questions, which I’m completely fine with ask having questions asked, uh, of me. I may
not have the answers and I’m completely realistic in that. Uh, the next slide is about why are we different? Uh, you
know, what what makes us eminent? Um, so we have kinds of intelligence that AI
doesn’t have. Um, and and this is really important because if we don’t recognize
what we have, we will give ourselves over to machines that will cease to
recognize us as being part of the equation. The current thinking around all of this is that these systems will
increase profit and they will drive out waste, which is a good intention.
However, underlying that is always the sense that uh in fact uh they will drive
out humans and since many of these systems are designed uh to deliver value
to humans at the end point um we have to be concerned about it. Now if you’re a
machineto-achine uh delivery system managing machines for machines own benefits all of this is
irrelevant and there is no human eminence except in the architecture. Um but the key thing is um going into
these points is you know consciousness uh of subjective experience. Humans are have a true awareness of what we need.
Um and it’s AI doesn’t have a knowledge of experience. It has past transactions.
It has a a transaction capability and a future transactions. But it doesn’t have
a relevance or a cross index to understand the meaning as they relate to
humans. Um we have um meaning
intentionality and meaning we have and I’m not going to read all the slides out because I put them up there for for the
purpose that you could read them. Um and I will make the the slide pack available. I’m not going to be um
controlling over it. Um the the point here is is that machines don’t understand human world. Why should they?
It’s not their world. They don’t uh have moral judgments. The uh Isaac Azimoff uh
laws or three laws don’t exist in reality. No one has applied the three laws to any of the systems. critical
systems that could cause death have a human injunction point that requires a
decision before a machine can proceed. Uh I think we’re moving forward now with military equipment that could actually
bypass that which is a bit of a concern. Um and the whole uh nature of uh machines
are that they don’t really understand creativity. They don’t they don’t create anything. they they re they repackage
um and they draw images from some places that they probably shouldn’t have. Um so
that that’s kind of an an idea around the human elements but the key points are that humans need to be involved in
this discussion around our world how it works and what is acceptable within it
because if they are not it won’t be acceptable and we’re going to be pushed
out of it. Um as solutions people, as people delivering solutions into uh
organizations, we have to think about how we are reorganizing them to function uh how
they will function uh better, but also what intrinsically
is made of the humans that are in that and what is machine. Uh and by having
that understanding, we can actually elevate the unique thinkers. we can elevate the people that understand the
end-to-end ecosystem. Machines don’t understand it. They operate it. And that
that’s really important thing. And machines are not uh infinitely adaptable. They will adapt within their
programming. Humans have the capacity to draw knowledge from multiple different areas to understand that knowledge and
to implement that knowledge into working practices and behaviors that actually make profit in a way that machines still
can’t. I’m not saying it will never get there, but we are a long way from it at the moment.
So, uh, AI can replicate a lot of human
outputs, but it doesn’t possess the underlying subjective, intentional, and value forming capacities that
characterize human minds. It just doesn’t exist yet. Uh, are people trying
to make it? Probably. I mean, I expect there are a number of mad scientists out
there trying to do this. Um, you have to wonder why. Um, certainly there is there
is the sense that some people never want to die and they want to propagate their minds into machines. Uh, I feel like I’m
in an episode of lawn mower man, but uh, still um, it doesn’t at the moment and we we
need to make some decisions around where humans should fit in the future world.
Uh, and at the moment no one seems to be really dealing with these issues. And then last um
the role of AI is to recommend, prioritize, predict, and optimize. We must maintain the final authority. I
saw something this week on uh LinkedIn that someone had posted an IBM statement from the 60s that you can’t sue an AI.
And I think that’s an essential point. So that’s as much as I’m going to share.
I’m happy to respond to questions.