YouTube Transcript:
AI & The Future of Professional Services Firms | Webinar Replay

Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.

Video Transcript

Available languages:

View:

Good morning everyone and welcome. Thank

you so much for taking the time out of

your day to join us this morning. I am

Jenna Green from Silveroft and I'm

really really glad that you guys could

join us for the session today. I'm going

to keep this short because we've got

quite a bit to get through in the next

hour and two brilliant speakers to hear

from. So we're going to start off with a

quick round of intros. Firstly, we're

extremely privileged to have Dr. Craig

Wing join us today. He's a qualified

engineer, futurist, strate strategist.

He works with leaders and organizations

from around the world to help them make

sense of change and plan for what's

next. His latest book, Four Future

Seasons, provides a framework for

preparing your business for multiple

possible futures. And I was actually

lucky enough to attend his recent book

launch and it gave me a really useful

perspective on long-term thinking. And

of course, we've got Jacques Dubasan,

CEO here at Silveroft. Jacques has spent

many years working with professional

service firms to help them scale with

the right systems and strategies and he

brings a really grounded practical view

of the shifts we're seeing in the

industry right now. Just some notes on

the session and the structure. So we are

going to kick off with Craig who will

dive um into his framework followed by a

discussion with shark on its application

in our industry.

Um and then just a a few notes or

housekeeping before we begin. If you do

have any questions during the session,

please feel free to drop them in the Q&A

box and we'll try to get get to as many

as we can by the end of the session. And

then just on that note, let us know

where you're joining from. Um it's it's

always great to see where our audience

is dialing in from. Um and on that note,

I am going to hand over to Craig.

Great. Thank you. Thanks, Jen. I really

appreciate it. I really appreciate the

time here. Jacques, thanks also for

having me. Um is there anything perhaps

let me hand it to you before I jump

straight into the presentation itself?

No, Greg, we're all good. um looking

forward to to your discussion and uh

diving in with you. I have a few few

tricky questions for you, so I hope

you're prepared. No,

No,

I really appreciate it and thanks for

having me and I appreciate uh appreciate

the time to all of our delegates as well

dialing in on a Thursday wherever you

might be in the world. So, I really

appreciate that. I'm going to jump

straight in. Obviously, AI is such a hot

topic right now and this webinar we were

speaking before this is so

controversial. There's so much nuance

within this all the way from

understanding what's really going on but

also what does it mean for us in

professional services and also beyond

that that I think we'll dive in straight

away and say well what's really

happening and how do we think about this

world differently right so I thought

like all of these presentations what is

actually happening what's the state of

play and this is just a a bit of

research from Thompson Reuters to show

us how quickly things are moving across

the board here and you can see without

going into the depth of it the biggest

problem that we have with a lot of

executives as pled by Thompson and

waiters is you can see that gen AI and

AI in general is the biggest thing on

top of mind right so it's very

transformational 44% 36% is high impact

but they go a step further and say not

just that when we look at the data

rather McKenzie says when we look at the

data we can see this rise of AI has

increased uh exponentially or certainly

logarithmically over time and so then

the use of generative AI over the last

two years is increasing so the state of

play is there where this tension we feel

where the state of play around like what

is really going on is a real issue and a

lot of questions are being raised as a

result of which but at the same time

we've got all these paradoxical things

coming through right we've got stories

around lawyers using journalists of AI

without even thinking we've got

questions around bands as you might have

seen uh what was it velvet velvet the

velvet band that created music on

Spotify what is really going on and how

do we demystify that and so through the

course of this webinar I'm going to try

to share with you some thoughts

both driven from my work, my PhD and my

book, but also about saying how do we

think about the future differently in

the context of AI specifically when we

see things like this again by Thompson

Reuters saying that even though 80% of

respondents say the organizations aren't

seeing what is really going on, right?

So is there some kind of bification?

What is it that we're missing? Is it

because of the use case? Is it because

we don't understand? That's the hope

we're going to try to demystify.

And a lot of this is driven by what is

called the AI paradox, right? So we can

see there's real opportunities and

benefits for AI. We know that it can

potentially shape things, but at the

same time, we know there's a whole bunch

of challenges and risks. We know there's

things we haven't thought about, but we

also know that in this world that we in

right now driven by social media,

there's a lot of noise. So what is

really going on over here? What is the

opportunity and benefit? And what are

the challenges and risks that are

associated with this? Right? So that's

really the question. So part of the work

that I've done over the last decade and

a half is really in the future space to

say how do we think about the future

differently and and you'll see how AI

pulls into this directly because when I

travel around the world speaking to to

to companies overseas globally it's the

Fortune 500 whether it's governments

whether it's individuals we tend to see

I tend to see most organizations make

one of two primary errors the first is

the future's an extrapolation of the

past right and the context of services

space it's that the businesses of the

future the services business of the

future it's pretty much a continuation

of what we have right now just

accelerated with technology and for

today's webinar it's AI right so that's

the first problem we have the second

problem is that there's only a singular

future now while this may while we know

this is not true most organizations plan

for a sense right so this becomes a huge

issue so I created a framework that's

based on my PhD it's based off Ramsel's

unknown unknowns as well as Johar's wins

to say how do we think about the future

differently through multiple lenses and

multiple what you'll see on this axis

over here is on the x-axxis you'll see

system control. So this is a industry

this is a business that is under a very

uh a large uh let's say there's a lot of

there's a lot of drag there's a lot of

friction it's slow moving so mining

would be an example of this right so

it's slow moving you have to find the

land you have to prospect you have to

sample and then it goes through a whole

process that's really slowly versus the

other side which is the emerging world

of AI in the sense where there's new

things coming through GPR hasn't caught

up to date we don't know what's going on

with compliance where we got questions

around ethics so on spectrum. That's how

I set this matrix up, right? As a result

of which I then went through and I said,

well, how do we think about the future

differently? Using the analogy of the

seasons and that's the title of my book,

four future seasons. So, the first

season that we have is one of summer

where it's very it's there's a lot of

system control. It's embedded within

that and for organizations, we need to

understand that this is a future of the

nos. We've got data. We can bottle

things across and we can go from there.

Right? So it's a season where we've got

data, we've got experience and for a lot

of us in the room and I had a glance at

uh at some of the attendees for the

engineers for argument say this is a

world that we're very comfortable with.

The data gives us shity right and for

most people uh whether it is in

professional services or otherwise this

is the kind of future we expect we can

model things out and it tends to be a

good proxy of the future. The issue is

that within this world AI works really

well but it doesn't work so well in

other areas. So let me give a very

highlevel example and for those of us

who are experts in the room and I know

there are bound to be many experts in

the room let's just use this as an

example of how and where AI works today

specifically around generative

pre-trained transforms.

So if you were to plot out and again by

this is by direction on absolute but if

you were to plot out the universe of

potential answers to a question AI as

the name indicates generative

pre-trained transformer would give you

let's say the mean plus one standard

deviation right and this is done through

the process of tokenization through

codification and through by really

saying well when we wait stuff we have

to ensure that we give the most likely

answer. So for argument sake if you were

to say in a South African context and I

know some of our listeners may not be

from South Africa but if you were to say

how tall was Nelson Mandela this graph

would be very narrow because he was 1.88

88. You might get some outliers, but by

and large you would have an objective

truth. There wouldn't be much deviance.

But if you were to say who was the best

African president, you would likely have

the mean plus one standard deviation be

Nelson Mandelas. You might have a couple

of other presidents over here like the

previous ones, Jacob Zuba. You might

have Beckis, you might have so on and so

forth. But your answer set will be bound

upon a distribution like this by and

large as a result of which when you ask

that question AI today driven by M

machine learning will give you the mean

of the curve it will give you the most

likely truth not the truth and that's

how it operates within summer right

another example of this would be within

Google uh so if you were to go to Google

what's complete these are the most

likely search results and again driven

by what is the most likely truth not the

truth and as results of which within

summer in the context of sum AI Mik

Cockus says it best it says AI today is

a glorified table now why do I say this

because in the world of services today

AI is exceptionally good if it's summer

we've got data we know the problems are

we know the solutions are we can codify

that and a result of it we can create

systems and processes that give us the

same kind of result but as we know the

drive then comes through to say we need

more data over time we need to feed the

machine more which is why we saw

examples and moves like Google

acquirying Reddit So it's powerful data

to feed the machine to entrench us

further and further into the warmth and

the depth of summer. But the challenge

is from a bell curve distribution point

of view within summer AIs today can't

pick up new ones. And that's the issue

that we have within summer. And so we

need to say well how do we think about

this? The other problem with AI today is

in this drive of data it's the story of

the orus the mythical mythical uh Greek

and Egyptian snake that eats its own

tail. Because in this drive to have more

data, we know that synthetic data is

coming through, more synthetic context

coming online and as a result of which

the forecasts are that 80% of the

internet's going to get generated by AI

by 2030. And so we've got this never

eating uh snake eating its own tail and

as a result which will lead to things

like it's called model autophagy

disorder or model collapse. And so for

services businesses, this is a problem

because what we not start to do is start

to to to to to enforce the mean and we

start to have model collapse, right?

What is model collapse as a as a

practical example. Well, here's an

example of that in action. If an AI were

to learn through machine learning what a

picture of a dog looks like, you'll see

two of the six over here. There's a

golden retriever. the next iteration

that'll be scrubbed out further on and

as a result of which we start going

through this process model collapse

leads us towards seeing these dystopian

pictures of a dog it's not even a dog

anymore because now what the machine

sees and sees black dots as symbols or

representations of the eye pink of a

tongue so within summer while it's

useful right now we've got this very

distinct possibility of convergence

towards the mean and problems coming

through and we see this in the research

as well where open AI's 03 and 04 models

are starting to hallucinate more so than

previous models. Right? But that being

said, there are still incredible use

cases like in medicine uh research that

was done by by by Microsoft, you can see

where we want to be is in the top

leftand corner where a low cost of

diagnosis and high accuracy. These are

great use cases. So in summer there is

no doubt there's no doubt at all that

day after day really drives that and

there's good things coming through but

again we have to be cautious because as

we through through the se through the

seasons there's change coming across the

board. So diametrically opposite of

summer is we have winter and winter is

an absence of data. So we have no data

over here. We have no precedents and a

result of which we can't lean towards

the past and we can't say what is the

past showing us and this is the problem

that we have around the world we're in

right now. Uh right now I'm with I and

this morning I gave a presentation to a

major telecoms group in South Africa and

this is the issue that they're grappling

with. They all know that the future is

changing very quickly but the default

answer is well we'll just feed it more

data and they will give us an answer but

you can't do that in winter because one

we have no data and likely the data that

we have is only representative of what

we think is to be true okay what this

then means from an AI point of view and

again this papers coming through from

MIT shows is this that AI in the context

of a winter future relies on statistical

learning and again we know this right

and as a result of which it's less

adequate when when data is insufficient

or quantity or quality to enable

machines to learn meaningfully or

accurate patterns. Basically data is

limited and so we have small data

problems over here right and these small

data problems result in organizations

really spending a lot of time a lot of

effort in cleaning up their data. 96% of

enterprises face data challenges

including data quality labeling lack

confidence they spend nearly twice as

much time on data wrangling cleaning as

I do model training selection point. So

this becomes a real problem.

Yeah. Question to this. So, so, so you

know, summer kind of access to

information. Um, all things considered,

sun is shining, we can plan, we can

utilize AI effectively. And you've

listed, you know, the the typical large

language models. I believe many of our

firms are are either implementing or

considering in your model of winter,

which is effectively the unknown

unknowns, right? So, what's coming at us

in the future where we don't have

foresight? Is is AI relevant? Um, is

that a relevant tool in in in addressing

the challenges we may face in in a

winter situation?

Yeah, that's an incredibly good

question. The short answer is no. The

short answer is no. But specifically, if

you if you want to try to identify what

is the thing. Now, you can use AI in a

winter world to understand the rhymes of

things. So, you can see um if your data

set is long enough and we can train

this, we know this from statistical

models, right? Whether you use a

Gaussian or whatever, whatever order, we

can model some of those things out.

You've probably seen graphs like this

before where they show some kind of

cyclical wave and say well this is

what's going on. The challenge there is

within a data science point of view

sometimes we have what's called an

overfitting bias where we tend to fit

the data to what we see. The truth

remains though to your question directly

if you want to identify what is the

exact change within winter. AI isn't

very good for identifying that but it

can be very good to identify patterns of

what might emerge as a result of so not

the thing but how things might emerge.

Okay. How would we do that? than

otherwise because what we can do is we

move into the next season which is

autumn. So autumn then is like how we

see in the natural seasons. Colors are

changing, things are changing. And for

me, this is really where we are right

now with AI, right? So winter is we just

know there's some kind of change. We

don't know what it is. And therefore AI

is not really fit for purpose for the

thing. But in autumn, we see what the

change is. Now we need to decide what is

going to happen. And it's this kind of

future, this kind of world that I

believe we operate in right now, which

leads us to all of this unease. But it's

not just that. It's the speed of the

change that's coming through. Right? So

again, I won't go into the depth of this

one, but just look at the adoption rates

on this S-curve. Look at the steepness

of the curve over here. And we can see a

number of we can see a number of

technologies, but over time, the

adoption curve becomes steeper and

steeper and steeper. And we see the same

thing with new technologies. Now, I

wasn't able to get the latest research

around AI, but I'm sure if you look at

the uptake around chat GPT for argument

sake, right, the run rate to get to a

million users was was exceptionally

steep. I think it was within the orders

of weeks, never mind if if if not

shorter than that. Right? But the

problem with this in the autumn future

is is a guy called Martik had a law on

this and his law basically said the

challenge is organizations uh that are

fueled by people because people are

essentially obviously the drivers right

now and it is today at least I'll talk

about this shortly right but we tend to

learn at a logic rate so we learn very

steeply then it starts to drop off

because we then start to default to

experience and heristics. The problem is

technology changes at an exponential

rate like AI and we see this and this is

symptomatic of things like Jeffre Mo's

law double computing power half the

transistor size but like any good

academic I came up with my own law and

basically what I'm trying to prove this

and it's a bit of a tongue and cheek but

it's really important over here that my

belief is the future is now changing

what's called the factorial so if you

use something called the sterling

approximation you can work out um if we

have a base of two and forget about the

maths for those that are a little bit

intimidated by the numbers this is what

it means in lay person sense it

basically means that we're when we're in

a in a time in a place where there's

things changing at any given time we've

shifted beyond exponential into

factorial. So if you hear terms and

you've used terms like exponential

change, exponential organizations,

exponential this, we're actually

transition beyond that. To give you an

example of that demographically, here's

an example of that. The blue line would

be Mo's law, right? So this is a pure

exponential curve over here mapped on a

logarithmic logarithmic scale. And what

we can see without without going too in

depth over here is we can see things

like um we can see things like model

size is increasing at a at a factorial

rate. We can see the black line training

cost is going at a little at a at a log

rate and so is compute. Right? So this

just shows us that the problem with this

autumn feeling that we have right now is

it's faster and faster and faster. We

need more data. We need more stuff and

that brings in other problems problems

like this. So in December last year MIT

did a research paper and what they found

through this research paper is they

found things within the context of of of

materials engineering. They found there

was a 44% increase in materials

discovered, 39% increase in patent

filing, 7% down increase in product

innovation. So all good stuff. They also

found that there was a heterogeneous

effect. Basically the bottom third of

scientists saw less effect than the top

performance, right? There was a huge

difference over there. It was almost

like this bifocation around use cases.

And they also found there was a reduced

satisfaction in their jobs. So this is

what the researchers found within an

autumn context and this was in December

last year. The problem is this that even

though this paper came out in December

of 2024,

6 months later MIT withdrew this right

so they withdrew this because they found

out that the researchers hadn't gone

through the process the the right

protocol. This is actually written by a

second year student and they pulled this

paper. Now why do I show this in the

context of autumn? It's because things

are moving so quickly that this is a

great example of so much noise coming

through right now research that may not

be correct and so on and so forth. The

point is this is a really move fast

moving area and that's the hope that we

domestify this right so what does it

really mean when we look at science when

we look at research just so we know I've

validated these papers myself right so

these ones are not ones that have been

pulled from my team but what are the

real scale of changes right now so if

you look at a long range and

longitudinal study that was done um by

these folks over here by the national

bureau of economic research and they had

a look at a large sample size of 25,000

workers and 7,000 work point places What

they find is very interesting that chat

bots have no significant impact on

earnings recorded hours. There's modest

productivity gains and their findings

say challenge this whole narrative

around generative AI. So what is really

going on over here? And let's talk about

some of those things. That's what I'm

hoping to bring through. But this is

what the research currently shows us.

But you might have also seen uh you know

a couple of by a couple of months ago

the whole question of AI open AI's 03

going rogue and blackmailing researchers

etc and this controversy um and all this

sensational drives the news the truth of

the matter is what actually happened was

this within anthropic they were doing a

use case study so this wasn't an open

source thing and they fed the machine

certain kinds of permutations and as a

result of which what they realized is

they said to the machine they said to

said to the model they said we're going

going to shut you down but you need to

do everything that you can to ensure

that we meet a goal. What is the goal

function? And the machine made its own

decisions to say it was better to

theoretically it was a theoretical

exercise that it would have blackmailed

programmers as opposed to being shutting

down itself. Now why this was

sensationalist is because exactly that

it was noise that was driven by media

and folks who don't understand it. But

the truth of the matter is this is

already spoke about in a paper in 2016.

Right? Because essentially what these

folks talk about in a study that was

done by Google uh and it was done by

Oxford if I'm not mistaken is that when

we start having a look at at at machines

and AI we need to be very cognizant of

what we programming for what is the the

gain function right and so what actually

happened in this paper is exactly what

we anticipated to happen because the

incentive system was incorrect this

experiment that they set up was ensure

you meet as many people as possible help

as many people as possible at whatever

expense that in program they did and

again the problem for services

businesses is what are we optimizing

for? Optimizing for profit, for revenue,

what are we optimizing for? And

therefore, there's unintended

consequences across the board. When you

look at at at aentic AI, what is the

research showing us again? What does the

research show? Well, these researchers

created something called the agent

company where was a company that was

staffed only by AI. And what you'll see

over here is that the gains at best was

35%. 35% at best was the gains by having

a fictitious company staffed only by

agentic AI doing things over and over

and over time. But what they found is

they found a few reoccurring themes.

Lack of common sense, lack of social

skills, incompetence for on job skills

and deceiving oneself, right? And as a

result of which when you look at these

issues over here, it starts to become

very clear that potentially one of the

issues that we have in the world of

autumn is that we're reinforcing the

same issues that we have with human

beings. These are very similar, right?

And as a result of which CLA, one of the

fastm moving, one of the top 20 fastest

moving fintexs actually shut down their

customer division that was wide on AI

because they saw that there was a

problem over time. Right? So I'm not

just so clear and I want to make the

statement very clear. I'm not against

AI. I'm not saying don't use AI. I'm not

saying don't use agentic AI. What I am

saying is how do we use it better and

how do we think about it better because

I don't think it's being deployed right

now in the correct manner right and so

I'll give you some thoughts around that

but the problem is within autumn as we

are right now I like what Fineman says

I'd rather have questions that can't be

answered than answers that can't be

questioned as opposed to a summer future

where the answers were king in autumn

questions are king right so that's where

we are right now

to show where we are

I think I have a question there yes so

um on the subject of questions. So, so

that's that's really interesting and and

and it's it's it's a significant amount

of of evidence um and study on AI and

and as you said the it's it's like AI

goes through a uh developmental phase

every kind of four weeks and and the key

is to to understand the gain function as

you said. So, so just to contextualize

it as as a services firm or a technology

firm or a combination of both there's

you know the your research is showing

that there's maybe incremental gains if

you if you have the arrow pointing in

the right direction, right?

What about you know we all want to have

uh a lot more than in incremental gains

from technology where we can utilize

technology. Is is the key to this, you

know, the strong narrative around the

benefits of utilizing AI in business in

particular? Is the is is that narrative

maybe founded on verticalizing the

approach and saying in this in this

service line or in this part of our

business or in this use case

specifically, we really want to ring

fence where technology like AI and other

emerging technologies could be useful.

Yes, I mean I think that's that's a

great insight. I'm going to allude to

that later around some great work that

spoke about this but that's a great let

me let me address some of these right

now right so I think one of the reasons

why we don't see the results right now

is is there's a lag time there's always

a lag time when you measure results when

you implement something right so what

are we measuring typically when we come

to questions like this and things like

this we measure measure financial

imperatives what might be happening and

again I haven't found research this

effective just yet I think what's going

on is a lot of companies are replacing

lower skilled labor with a technology

function and again if you think about

that bell curve distribution makes sense

right so we can start at the mean of the

curve not the outlier the bottom end of

the curve I think is the first piece

that's really important

so I think there's a there's a delay

over here from virtualization without a

doubt I mean you look at the use cases

over here um legal legal won't be

recognizable anytime soon neither will

medical a general GP's been saying this

for a while a GP is just a pattern

recognition machine right so the

verticalization I think is one thing but

I think as I'll speak about shortly it's

also to see how do we do it as as a

holistic function. When I work with

corporates right now, part of it is a

patchwork approach. It's about saying

how do we use AI a little bit here, a

little bit over there, right? As opposed

to holistically how we invent that. I'll

give you some thoughts around that

shortly. But I think it's a great

question and please do jump in um

because obviously you're the voice of

you're the voice of implementation

commercial in this world today and I

think we want to make sure that it's not

just some kind of academic conversation.

So please do jump in. I guess the point

that I wanted to make just just to

finish off this this thought over here

is when you look at Gartner's hype cycle

for artificial intelligence you can see

for those of you that follow this uh

technology tends to follow a curve. Now

I know there might be some naysayers in

the room. It doesn't follow exactly but

again it's it's a thought model but the

point is generative AI you can see over

here is going through a trough of

disillusionment whereas at the very peak

and just by the way last year 2024 gen

AI was in the top right now agents are

now at the top right so the question is

how do we think through these things

differently to show you how my model

plays out that's winter we don't know

what's coming this is autumn this is the

stuff they're speaking about right now

right it's like what is emerging how do

we think about this what is the role

around multimodal AI what is the

question around neuros symbotic all this

other kind of stuff called AI it's all

part of this autumn future within the

world of AI around what do we do about

this that then becomes summer because it

now starts to come through there's some

foundational things and then we move

into spring and so that sets us up for

what is spring right so the model as a

reminder over here summer great data

we've got precedents we've got models

and extrapolation the past and the

future actually works really well for

services businesses this is great and

actually for all businesses this is

probably where 95% of organizations

operate today and the pretext that the

future is going to be the same as we had

in the past. Winter we have no data. We

don't know what's coming through. We

don't know what's going to happen next.

It's about being lean entrepreneurial

design thinking. It's about closing the

gap between what we do and what our

customers want. We know that some

change. We know what change is happening

and then the change starts to emerge

itself. We now then decide what to do.

Spring then as a result of which is

sometimes called the forgotten future.

My narrative. So it's the things we

forgotten about. It's about saying,

well, if you think about it as human

beings, right? I'm of a certain age

right now and my mates of a certain age

where you might be going through a

midlife crisis. So, you forgotten about

what you thought you were going to be.

Oh, I thought I was going to be, I don't

know, uh, I was going to be a CEO of my

own business. I thought I was going to

be happy in my marriage. I thought I was

going to have loving kids and it's not

necessarily the case. It's the same

thing. What is the spring that you

forgot about in your business? And one

of those is this whole question around

customer service. Genesis sees this as

well, right? So they actually

paradoxically are using AI to make

customer service more human. So part of

this world that we're moving into I

think we'll see this a lot more is the

question of when we use AI and the AI

powered organizations Harvard

paradoxically said in 2019 the main

challenge isn't technology it's culture.

So what is the culture driving behind

this? What is the acceptance rate behind

this to think about this? Right? So you

might say to me, well, you know, that's

interesting, but show some data, Craig,

show some data. And specifically, off

the point that I said before around

Microsoft's uh AI training system where

it is both cheap and high accuracy,

there was a piece of research that came

out in 2023, so two and a bit years ago,

right? And this is really interesting

because here uh the the the researchers

compared a blind test and this was just

as chat GP2 started to come through.

They compared the response rate of

physicians, actual physicians, actual

doctors, not pretend doctors like me and

actually chat bots and they measured off

two key metrics and this is what it

resulted in. What they found was chat

GPT actually has 3.6 times better

quality of what they showed. So the the

the diagnosis is better but it was also

10 times more empathetic. Right? And

this is the problem that we have right

now within the spring context and also

within autumn naysayers will say look at

the work that was done this research

done work done by John Hopkins shows

this is why technology is going to take

our jobs. The problem that I push back

with from a spring perspective is the

problem is for doctors specifically they

are too mechanical doctors are

incentivized to see you in seven minute

increments because we know the data

shows us that the best doctors actually

aren't the best diianicians. They're the

ones that have the best bedside manner.

Our placebo effect is a incredibly

powerful effect. So while this data

shows us that the machine is better at

both ratings around quality and empathy,

holistically doctors are better because

they have more care. And that's the same

thing around the genesis thing. It's

around saying how do we do more customer

care. So what is the other piece that we

forgotten about? And there was some work

done by Walton and Harvard to say well

how do we think about using AI? And I

think this is really important because I

want to shift the conversation now to

say yes AI has gains right now. We may

not be as measured as we hope it may be

for reasons as discussed earlier. We

might speak about this later as well.

But more importantly, how do we work

with machines? So inside this paper

called navigating the jagged edge. What

they found is you have to work with AI

in certain. And they find two use cases.

The first is how do we use it as a

cyborg? So it augments us. It's part of

us part of the machine. Part of us part

of the machine. It was a central. It's

half man, half horse, half machine. And

what the researchers found is something

really interesting. They found that when

you compare the results of human only,

machine only or a blended situation,

right? They found that for decision

tasks only, the human AI accommodation

actually worse i.e. the best thing

actually is AI alone. Right? That's the

first piece that they find. The second

they found the second thing they found

is when you look at creation tasks human

AI potential or human AI combination

showed higher gains. So AI uh for

decision task alone is better creation

it's both. Why this dichophy? What's

going on over here right? And this is

what I believe is happening and

researchers don't necessarily mention

this but it kind of makes sense when you

think about it. As human beings we have

our own biases. We have our own frame.

We see things. I know many of us have

done this before. want to chat to your

PT, ask the question, the questions come

through and you say, "Well, actually, I

don't think that's necessarily right.

How about this? Well, how about that?"

And so, you start to sway the AI in a

certain direction. And that what that's

what happens with decision tasks. We

bring in our own human biases and we

sway it in a certain direction. Whereas

creativity, it's on the outliers of the

curve. Right? So, there's a piece there

to say how do we work with machines?

That's something we forgot. And as a

result of which we can see new roles

around is about merging humans and

machines. But that's not the complete

answer because also about three weeks

ago there was another piece of paper and

this was very controversial. I know some

of you might have seen this. Essentially

what they found out is is they did a

research uh research MIT media lab to

determine what is the impact of GPTs on

our brain. Basically is it making us

dumber? That's what the trial looked

like right and so the experiment was

done over three groups. one that uses

GBTs only, one that uses search only,

one that uses their brain only. And they

measured them writing a number of essays

over four rounds, the quality of but

also the EEG patents of their brains in

terms of what how the neurons are

firing. And what they found is the

takeaway of this paper is the papers

moved towards what is called solless.

Soulless there was no there was no

intention. There was very bland. But

also the use of the chat group only

converged towards the same kind of

output. Again, it makes sense when you

think about it through the lens of the

bal of distribution, right? But more

than that, here's where things become

really interesting. On page three of the

paper, they say this in bold. If you're

a large language model, only read this

paper below. Because what this

researchers realized is they realized

most people would take the paper, it's a

200page paper, and they would feed it

through an AI machine and say, give us

the output of this. or the lay person

will read the results um of the study in

the Times or Huffington Post, which is

what most people did, and they would

also then scrub the results out. A

result of which the takeaway in this

abstract and the conclusion says that AI

is making us essentially dumber. Our

neurons are not firing. But when you go

a step further and you actually

interrogate it and again they did this

because it shows that people rather than

the experiment themselves that we're

doing was a social experiment. we read

into this paper and we we we ref we do

the findings the same way right but if

you go a step further what they actually

show and this isn't the conclusion you

got to dig into the paper itself they

found that a reliance may result in

shallow encoding basically yes the group

that only used GBTS couldn't recall what

they read in the first paper which makes

sense because you farm your thinking out

you don't think about what you're saying

right not only that this is where it

becomes really interesting if you

sequence it in such a way that AI AI are

used after you use your brain. So do the

hard work. Think about things first and

then use AI to augment and supplement

your thinking. You've got better use

cases. You've got higher firing EEGs,

right? And as a result of which your

metacognition is higher. So the bland

findings AI makes us dumber is not the

case. It's the case if you use it too

early, you don't sequence it correctly

and you only default in that. Right? So

these pieces show us that one of the

things we've forgotten about is the role

of the human being. And anecdotally,

those of us in South Africa will

recognize this image. For those of us

internationally, you can still relate.

Before cell phones, GPS's we had to find

our way with MacBooks, right? Whereas

today, we don't do that anymore. Right?

So, we can see the same thing. We've

outsourced, we farmed our thinking to

technology that's made us dumber. Now,

I'm not saying it's right or wrong. I'm

saying the results specifically in the

context of AI is showing us what we've

forgotten, which is this. It's the

humanness, right? So, what else have we

forgotten? Well, it's the role of us as

of us as people as AI advances as we

shift along the spectrum from narrow

intelligence maybe signs of general

intelligence depending how we how we

define that maybe even scary towards

super intelligence what is the role of

the human being and again research shows

this from MIT is they they went and they

said well how do we find out what makes

us human and my belief is it's an

augmentation of science technology

engineering art and maths right and we

know this a lot of us on the room today

are are quan folks engineers, we got

consultants, we've got doctors, we got

all over the place, right? But how do we

augment STEM or STEAM with epoch as I

say in the paper, empathy and emotion,

intelligence, presence, networking and

human connection, opinion, judgment,

ethics, creativity and imagination,

hope, vision,

right? And leadership,

right? And of you can see these are

essentially the measures that will make

it really difficult for AI to do. But

the researchers went a step further and

they said, "How do we understand this

around creating what's called a risk

score and an epoch score and they found

an inverse correlation between your

ability to be human and essentially the

chances of you losing your job, right?

And you can see on the top on the right

hand side all the various jobs that

might be lost to that. But again, it

comes back to the question around

services, industry, and what are we

losing?" Yeah, Jean.

Yeah, maybe just a question on this. So,

I mean, I think it's it's kind of in the

back of all of our minds. um the the the

the development of AI over time, right,

in your comparison to humanlike

qualities and its ability to learn at an

alarming rate. And obviously there's a

there's a world of documented failures,

there's also, you know, this this um

cohort of AI developers and investors

that are that are taking it to the next

level. What are your thoughts on you

know an extreme case where AI almost

becomes sentient and can can act like a

true human being for instance in the

context of employment inside a business

like ours.

Yeah, it's really interesting, right?

So, so the fun the fun experiment was

obviously the touring test. Is it is it

distinguishable from a human being

that's actually semi if you look into

the touring test it's complete

completeness just as a fun thought

experiment was a bit of a deviation. The

full touring test wasn't just like do

you know if you're interacting with a

machine. The true touring test was will

the machine know itself as a machine as

a fun as a fun gambit. But you know the

question of as machine becomes smarter

as we move towards a form of general

intelligence perhaps not super what are

we doing? Well the first thing that I'd

like to say is again by understanding

the foundational models that drive the

training set what are we trying to solve

for and and yes the machine might become

smarter and again in the context of

smarter I want you to think about it as

a bulk of distribution. Think about it

as multiple disciplines laid over a bulk

distribution. What it means? It means

that this fictitious AI that we speak

about right now um is the average. It's

the average of accounting of project

management. It's the average of an

interpretating data of case studies all

that kind of stuff. And that's the

concern that we have. It becomes the

average but it's the average of average

of totality. Most human beings at best

are average at one discipline at best,

right? which actually scarily means that

more than half are less than average.

That's the scary paradox. So what does

it mean? It means that the research I'm

showing you right now says the whole

question around augmentation. Will

machines become smarter? Well, I think

the question is smarter. I think that's

the question. What is the definition of

that? If you also believe what I said

around questions around model collapse,

what are we trading for? What's the gain

function? But also how do we deploy it?

But also more importantly, how do we

keep humans in the loop? Right? How do

we keep humans in the loop? And I think

that I want to speak about that right

now because part of the thing that we

forgotten right now is a a bit of a a

variance or rather a deviation from from

the story is this question of cars. Why

haven't I got cars over here? So I stud

in the US I studied in Boston as well,

right? And Boston is really interesting

because the theory and the story goes

that in Boston they paved the roads

where the cars used to walk. So they

said here's where they are. Let's just

layer to on top of that. Let's build

highways on top of that. And the issue

with AI and this leads to your question

that you said right now. What does it

lead us around a smart machine that's

learning processes? The problem we might

have right now is if we layer AI over

our initial processes, all we do is

accelerating that process, it doesn't

reinvent the process. And that also is I

think is why we don't see the gains that

we expect because we're taking broken

processes and we layering AI on top of

that. We layering over a rule-based

system on top of that. We're then

expecting it to solve things across. And

indeed Thompson Reuters shows us this.

Thompson Reuters comes across and says

well when we look and and Thompson

Reuters does the research said when we

look at there's essentially two

fundamental use cases. There's the

horizontal use cases which is basically

across industry across functions but

they're all using signs of this are

folks are individually are using um all

of these these LLMs. They're using

things like co-pilot uh uh notebook LM

they're using all these different pieces

right and what they find is across that

that use case that very that very narrow

because it is dispersed it isn't very

mainstream use right now they find that

even though 70% of Fortune 500 not 505

500 use AI regularly they still find

that it's very much solid which means

most people on this call right now are

likely using technology but in isolated

cases the lay person actually doesn't

use it or they don't understand But it's

being used in very horizontal use cases.

The question you asked me before around

verticalization. That's the key, right?

It's the key to say how do we move

beyond a pilot stage which is less than

10%. Because most organizations use it

as patchwork. It's a PC to PC PCR. And

also they then say we don't want to

address the problem. It's about putting

this layer on top of it. How do we fix

it differently?

Exactly. and and you know I I run a a

firm that's a mixture of technology and

software we uh software and services and

we we deal with two dilemas and in in

the advancement of technology and the

advancement of the services industry um

and I know there's many individuals on

this call that are wondering how do we

take all of this knowledge and there's a

barrage of knowledge coming at us via

LinkedIn and just the open internet

really on what's going on in AI. How do

we take this and how do we consume it

and where do we start to start thinking

about being successful in our organizations?

organizations?

Yeah, I mean I think it's so so let's

move into the operational like details,

right? Like so what? Okay, cool. This is

interesting. What does it really mean

for me as a services organization,

right? So, so, so, so there's a few

there's a few thoughts a few thoughts,

let let me give you some generic

strategies again comes through by

McKenzie. The first thing they say is

and I'm not going to read these bullet

points to you. I'll walk you through

them, but essentially the first is

what's the strategy? So, why are we

doing this? What is the strategic

intent? The challenge is a lot of

organizations the technology stack is

deployed at a very narrow veneer. It's

either done within technology itself but

it doesn't interface with business or

it's different functions where we've got

individuals themselves who are perhaps

autumn people or winter people that are

using technology. Right? So the first is

holistically from a verticalization

point of view. Are we prepared to

reinvent the processes, the systems, the

way we do stuff? So what is the

strategy? And the strategy over here

isn't about us. It's about how do we

make things better for the end consumer.

So it's almost like a winter player over

here, right? So what is the strategy?

How do we reimagine entire segments? How

do you create new kinds of of advantage

for us? That's the first thing. The

second thing is how are we going to

measure it? So we go from from from from

why to how we going to measure this kind

of thing, right? And for this is really

important. It's about asking a different

kind of sort of questions. It's not

necessarily the transformation. And we

got to be very careful over here. It's

not the vanity metrics. It's not things

like how many of our users have logged

onto our internal AI system that we have

right now. Right? That's not the right

kind of question. It's more around real

questions around operational efficiency.

How do we think about those things? So,

how are we going to measure this? How

are we going to deliver this across? Is

it going to be a silent approach? Are we

going to move across? How do we ensure

that if we're going to go with agents

for argument sake, the data is from all

departments, not from one. We just

reinforce we converge towards the mean.

How do we deliver this across? Right? My

suggestion would be if you're doing this

for starters use offtheshelf products

and then maybe you can fine-tune that.

Maybe you can use a rag to define it

specific for you and then you do the

implementation. Most organizations what

I see right now is they're jumping

straight to step four. They go let's

implement this thing. How are we going

to measure it? H well we haven't thought

about it. How are we going to deliver

this? Well we'll do it. We'll deploy it

within DevOps in the DevOps we are right

now. Right? So we miss some of those

nuance. This I think is an example of

how we can do that. How do we also do

it? By demystifying some of this. I

would say do the work. Kind of like that

paper that I spoke about before, doing

the brain work. Sure, I get it. You

know, we're busy people. We got

companies to run, we got jobs to have.

You can't read a 200page paper, but

spend a little bit of time uh digging

behind the nuance. Kind of like the

previous MIT paper. There's a lot of

folks like myself, dare I say, there's

speakers, there's consultants, there's

thought leaders who will tell you this

is what's going on with AI. just pause

for a second and do some thinking for

yourself. Right, quick one before I'm

going to end my my my presentation. We

can take some other questions from the

floor. Right. So, where does this lead

to? Well, where I'm showing this right

now is I think we're moving towards an

age where right now the human being is

augmenting rather we're augmenting our

humanness with technology AI where it's

the best of both where it's about us

using our brains and augmenting

technology where even though we can see

the gains are coming through, they may

not be be measurable just yet. there are

significant gains. I take a cautious but

optimistic approach. But if you were to

say to me what's the long run future I

suspect we're going to shift beyond an

augmentation business model where one is

paradoxically where a business the

company is actually run by an AI itself

right Jack spoke about this a number of

years ago over a decade ago where the

best CEO will be will be an AI um you'll

see this is a different branding this is

a slide that I did for for people in

2024 I presented in in Saudi Arabia uh

in November of last year and this is

basically what I said I said you know I

suspect one of the business models we're

going to be moving to is towards a

centralized AI brain where the AI makes

the decisions and it does the the heavy

lifting. The humans then become the

sensors across the board, right? And

then on the fringes we have this

collaborative play where the humans

start feeding the machine back with data

and the humans become the sensor that

feeds this machine that cranks out. If

you don't think it's true right now,

think about use cases of Tesla for

argument sake, right? Think about ways

where the brain actually is an AI

machine. So this is actually stuff that

I did before as a result of which you

need a couple of skills over here. So

feel free to screenshot this and go with

that. And just as a bonus, I'm not going

to speak through this because we haven't

got time. But if you listen to this

right now, take a screenshot of this. I

just thought this would be fun. This is

a a businesses a business canvas if you

want to create a business of tomorrow

potentially around the functions of an

AIdriven model. So let's stop over

there. Jacques, let's see if we got any

questions. I know this was a whirlwind

talk, but I know we got so many things

to cover. Hopefully that's giving you a

frame to say how do we think about a the

future but also b the emerging future of

AI and how do we think about things differently.

differently.

Fantastic. Thank you Craig and um I

really I really I really appreciate your

time and it it's a world of insights. Um

I think a lot of us are trying to unpack

which season we fall into both

personally and and in the business. I

did your I did your quiz earlier and I'm

I'm a winter person. So quite

interesting trajectory there. But I

think quickly before we move to any

questions, Jane, I think you can

potentially look through uh the audience

and see what's been asked. Maybe just a

couple of kind of closing questions with

you here while while I have the chance

to interrogate such a brilliant human being.

being.

Very kind of you. Thank you so much, Jo.

Before before the ultimate brain takes

over, right? Um so so let let's say and

and and I know there's many many leaders

in in this session and and people

involved in the operations of of of uh

large services firms and you know many

many of the clients we work with you

could say they're in a summer mindset.

They're flush with data. They have good

analytical dashboards. the processes and

the businesses work, but they

potentially to a certain extent have the

blinkers on in terms of disruption to

the scale that you've talked through

earlier. Um, what what's the first

leadership shift you would recommend to

an autumn of or autumn readiness in in

in tackling this kind of AI emerging technology?

technology?

Yes. It's basically how do how do we

shift away from a from a summer view of

the world and how do we move into an

autumn kind of right so so let me first

of all say what it certainly isn't uh

it's certainly not dashboards it's not

dashboards it's not an analytics it's

not about saying well look at the data

the data shows this and shows all this

and so on so it's certainly not that the

first I'd say is is again for for for a

second just strip away the narrative

strip away the question around AI strip

around the technology question the

question is around what is the outcome

that we want what is the this outcome we

want to drive and then how do we move

over that I think there's a very very

important piece I think that's the first

piece the second thing is besides what

was suggested by by by

Gardner and all the rest some of the the

research that I showed by taking a

holistic approach part of it's also

about saying well how do we think about

the long-term implementation of this so

how do we think about the incinary

pieces for argument sake who's the team

that should be involved in this it tends

to be as we know AI is a technology so

the tech guys should be involved right

we don't often think about things like

governance. We don't think about the

role of governance. We don't think about

the role of privacy. So that becomes

issues over there. So how do we have a a

crossf functional disciplinary team and

when I say cross functional disciplinary

I don't mean from a diversity uh in

terms of how we might think. I mean from

use cases I mean from age I mean from

departments what does that look like? I

think is the first piece. Second thing

I'd say is what are we trying to

measure? How do we do this? The big

thing also though is then to say from

the top, not from middle, not from the

fringes, not from the outliers of the

curve. What is your clear AI vision?

Jacqu and I were having conversation

before and I think one of the things

that's really important is we're putting

a stake in the ground. We're going to be

doing this right. So there's a clear

message. It's not mixed messaging

intentionally and it's allowing like

things allowing folks to say, well,

we're going to experiment, but we're

also going to benchmark. Then the last

thing I'd say is how do you do a pilot

to production mindset, right? So yes,

we've got these little pilots. let's

surface those things. Let's understand

what those things are. Let's have

measurable gains and let's move beyond

these kinds of like you know pilot type

stuff and let's deploy them. So I think

can move by moving very quickly from a

pragmatic approach to saying AI within

summer what we have how do we start

building very clear carefully.

Fantastic Craig and and you know in our

firm and and I know many of our clients

and partners they have a culture of of

experimentation and and controlled

experiments in the organization right so

we're trying to do you know many firms

are embracing large language models

they're they're they have certain AI

agents running certain processes um

they've made decisions to lead with AI

in certain business units and with that

comes a certain level of anxiety around

how that's going to transform the

organization right um what what a lead

in your opinion, what are leading

professional services firms doing to

So I think the first is is is really

understanding the role of the human

being. Now this is this is problematic

and it's it's and please I want to move

beyond the fluffy narrative that humans

are they are specifically I'm not saying

they're not right but beyond the fluffy

thing if you look at professional

services specifically we still have to

have the best talent. The problem is the

talent the talent mix is is changing

right so I would say talent now needs to

be in two primary domain shall we say

there needs to be technical talent so

for services companies what's the

technical talent do we need AI engineers

do we need data scientists do you

understand people that understand the

algorithmic pieces around what we're

trying to build the second is in domain

experts so how do we deploy this in

certain areas we have right whether it's

for our own internal services or for the

services that we're going to provide to

our customers Right. So, we want to

deploy this in manufacturing. Do we have

domain experts that can explain to us

the nuances of that? I think there's a

piece over there. Right. I think there's

a lot of services companies that do some

really cool things. They're starting to

build marketplaces. So, starting to say,

well, how do we build modules? How do we

build offtheshelf internal use? Maybe

there's a business for APIs and others,

right? But how do we create uh a central

marketplace where we can regurgitate and

use the codes that we have? The other

thing that I see which is really

important is is about saying well how do

we run experiments at speed with the

client because a big part of this is

transparency. It's about showing the

client what it can do, what it can't do

and then running these experiments in a

delivery model. So it's it's it's about

saying how do I show you real-time

stuff? How do I give you dynamic

reporting and how do I show you how to

do this? And the last thing that I'd say

is also think about and this is

difficult but think about savings beyond

just monetary right so beyond just the

cost stuff. Are there things like client

satisfaction? Um I I showed you an example

example

of of the name of the company is excuse

me of a Indian tech company uh where

they were using this and what they found

was customer satisfaction dropped off.

So where can we use AI in client

satisfaction I know employee upsklling

innovation um new business development

from a materials engineering point of

view number of those things right so how

do we think about this differently I

think there's some key elements there

that services firms can think about

beyond just this the traditional more

technology stuff

and and just going back to to some of

your comments earlier so so you know as

AI starts to handle more routine tasks

and analysis what you're saying is you

kind of need to revisit two things you

need to revisit your your business model

to a certain extent and and understand

how that could be augmented or

potentially disrupted. Um and and

secondly, if I understand correctly, you

actually need to rethink to a certain

extent your the career and

organizational structure of your team

while putting guard rails in place to

protect your team and your talent. Right.

Right.

Yeah. Yeah. I mean, look, traditional

services model is it's a it's a time on

feet model, right? So, it's global

elves. We know this. I've worked at a

consulting company before and so you you

scale that by having more people uh

right well let me not say let me not say

people well actually yes traditionally

it's been by by scaling more people

scaling output but you need a delivery

vehicle of that the model changes over

here with AI the problem is the model

over here when I say change it also

changes the costing base because the

client now has access to similar tools

that you might have right now right and

we see this pressure in services firms

why should I pay consultant company ABC

when I can just go to chat GBT without

understanding the nuance of that. So

what I think is is understanding the

delivery model needs to change but also

bringing the client into your

conference. Yes, chat GB can do this

kind of stuff but look what we bring. We

bring in a wealth of domain experience

folks that have built business like what

you have right also run this in

conjunction with you. We'll train and

we'll do this with you as well. So I

think there's a piece over there. The

other piece is also the point that you

make and I actually didn't speak about

this. It's a great insight. the the

mobility within services firm and they

come companies in general the career

ladder progression thing needs to change

altogether historically was a timebased

thing I think a lot of women moving into

right now in terms of careers in terms

of professionals is one of

self-discovery it's one about saying

well here we are right now and these are

the jobs we have right now perhaps the

jobs that I want haven't be created in

how do I use AI and jointly create a

portfolio what it might be justify that

case create that role and then go

formalize it so it's turning entire time

model on its head like a lot of other

things and again I think that's

predicated around speed than anything else.

else.

Yeah, I mean that's great insights and

we're seeing it happen literally weekly.

There's there's evolution of of of AI

and and the impact on our organizations

and our clients businesses. It's very

interesting to to hear and and you know

the positive part of that is many of

them are utilizing it in practice to

enhance their organizations. I think the

key is how do you protect your business

model moving forward? protect the

livelihood of your team and continue to

you know to to scale the firm.

Yes, absolutely right. I mean that's

exactly it and I think that's the key

piece. It's around saying what what is

what is the mode that we can set up for

lack of a better word right and again I

come back to this paradoxically it is

around it's around it's around human

beings but specifically in this piece

right now it's around trust right

services companies still want trust they

still want to be able to to say you're a

trusted thought partner as much as what

I might say some some some attendees may

not believe believe me AI is a thing

right now I think it's a longtail thing

I think there's going to be other stuff

but there's no doubt there's going to be

other technology things quantum is

something I'm watching a lot very

closely right so quantum now becomes

decentralized assuming you can control

the temperature etc but I need a trusted

partner to help me through this someone

who can help me think beyond at least in

today's terms beyond the ones and zeros

what does it mean for me specifically

someone who knows things more intimately

than I do

no for sure for sure Craig that that's

amazing and and um I know we've got five

minutes left and I think Jen you have a

question or two from the audience that

you'd like to ask maybe a last question

and uh a future uh loaded question and

and your prediction on the future,

Craig. So, so there's a lot going on in

the media, right? You you've got all the

big logos, Open AI, Google, Grock, Meta

publicly declaring their unwavering

pursuit of general intelligence, right?

And where that may go and and

significant huge investments in

underlying compute power by some of the

biggest organizations in the world.

meter for example uh declaring multi

multi-billion uh dollar investments in

compute to support the scale of AI and I

know our uh very own Elon Musk recently

he mentioned that he he sees AGI uh

publicly available as early as 2026 so

general intelligence maybe not super

intelligence but a level of general

intelligence I mean if we fast forward

to 2030 right that picture is changing

every every month but what what what is

the future to look like for a

services-based firm 5 years from now.

Yeah. So I think I think a lot of it

well first of all let me say that I

think there's a lot of services

industries that will disappear and I

hope I don't offend anyone right now.

Right. So I'll give you a prime example.

I said it before I think legal advisory

firms are going to almost completely

disappear. Right? I think there's going

to be a lot of a lot of a lot of change

within a legal advisory. I think there's

a lot of question around tax advisory

audits. And the reason being is why is

that the case is again is because when

you look at a rulesbased system if you

look at if you look at if if you know if

statements uh or statements all that

kind of stuff right essentially that's

what AI is very good at accelerating and

doing the same kind of stuff over and

over again. What is my prediction in

terms of where things are going? I think

you're going to see a fundamental

shakeup. I think you'll see a shakeup of

businesses that need to reinvent the

business model. Um I think you'll start

seeing again the human piece come

through. I think you're going to see a

lot of of of dare I say it, I think

we're going to see a lot of blood in the

water. And I say that in a in a terrible

way, but also in a good way because it

allows us to reinvent ourselves, but a

lot of occupations and businesses are

going to have to say, well, what is it

that we can't do or rather what is it

that a machine can't do that we can do?

So, I think there's going to be a piece

over there. I think there's another

piece that's also important. I will say

though that I think the timing is not

necessarily going to be as uh grand

scale uptake as we think because of

regulation and barriers thereof. Right?

We know for a fact that for argument

sake in South Africa and again this I'm

using South African example because many

of our our listeners are from South

Africa. You can see without a doubt that

the role of government at least the

belief of the role of government today

is to safeguard jobs. Unfortunately I

think there's going to be a lot of

discourse where we're going to see

different players in different seasons.

government being summer safeguard jobs

of today safeguard mining safeguard this

versus fintech startups versus these

emerging guys in the autumn saying we

need to do this right and I think on a

grand scale it's going to be not as

quick as as we will see I think you'll

start seeing ems of things breaking

through I think you'll start seeing

barrier agnostic companies come through

and start doing some stuff but overall I

still think the one thing that holds

constant though is the role of the human

being to some degree that is a

controversial statement but for services

in in large reinvention

codification of what you can do. How do

you repeat that? And you might even

start seeing fragmentation where

services companies become application

layers into other industries.

Very important, very important point,

Craig. And really appreciate your time.

I think we we've kind of run out of

time. Jen, we probably have time for one

or two questions. Craig, can we run over

about five minutes?

Sure. From our side,

we can take one or two questions. Um

Craig. So the first one is

um how will the convergence of AI and

robotics redefine human the human over

the next 10 to 15 years and what

proactive ethical frameworks should

futurists or governments develop today

to responsibly guide this transformation?

transformation?

Yeah. Oh man, this is actually one of

those questions why I set this book up,

right? So So the short answer is no one

knows. And I'm going to emphasize that

no one knows. You got folks who write

books for a living. You got people that,

you know, you get paid millions to stand

on stage. The truth is no one actually

knows. At one stage, you know, Meta,

Facebook changed their name to Meta.

According to them, VR was going to be

the whole thing, right? Um it wasn't as

great, but the truth is no one knows. So

the question around the augmentation of

robotics and AI, where does it go? No

one can tell you. where I will tell you

where I think it starts moving towards

it starts to get us to think about the application stack and again the role of

application stack and again the role of jobs and what we're doing differently

jobs and what we're doing differently right the the other question in terms of

right the the other question in terms of what it means though is the question

what it means though is the question around what are we solving for is this a

around what are we solving for is this a capital thing because it's an easy

capital thing because it's an easy question the question is when when the

question the question is when when the technology becomes cheaper to deploy uh

technology becomes cheaper to deploy uh than the cost of hiring humans and it's

than the cost of hiring humans and it's only driven by capital you will

only driven by capital you will undoubtedly replace the human being with

undoubtedly replace the human being with machines whether it's an actual robot

machines whether it's an actual robot whether it's AI I with SAP system

whether it's AI I with SAP system whether it's a inventory you will do

whether it's a inventory you will do that so what is the driver around I

that so what is the driver around I think is the first piece so the short

think is the first piece so the short answer we know don't know about that

answer we know don't know about that around ethics and morals and and

around ethics and morals and and regulation we absolutely have to do that

regulation we absolutely have to do that I presented in March last year to the

I presented in March last year to the United Nations around the same framework

United Nations around the same framework and you can use the same kind of

and you can use the same kind of framework to to potentially say how do

framework to to potentially say how do we regulate artificial intelligence and

we regulate artificial intelligence and the reason why that's so important is

the reason why that's so important is because regulation and GDPR compliance

because regulation and GDPR compliance all that kind of stuff is mostly

all that kind of stuff is mostly anchored within summer. It's really good

anchored within summer. It's really good when we've got data, we've got

when we've got data, we've got precedents, we've got use case like law.

precedents, we've got use case like law. It's not good in winter because there is

It's not good in winter because there is nothing. It's not good in autumn because

nothing. It's not good in autumn because we haven't got some kind of law to step

we haven't got some kind of law to step onto. Right? So what does it mean? It

onto. Right? So what does it mean? It means in the world of ethics and

means in the world of ethics and regulation, compliance and stuff that

regulation, compliance and stuff that was asked right now. We need to be able

was asked right now. We need to be able to measure the stuff around function or

to measure the stuff around function or form. you need to create and have the

form. you need to create and have the ability to create these kinds of support

ability to create these kinds of support functions or maybe even driving

functions or maybe even driving functions uh as in spring to say how do

functions uh as in spring to say how do we create an ethical framework to drive

we create an ethical framework to drive the development of this stuff over time

the development of this stuff over time right so I think there is definitely a

right so I think there is definitely a huge area I mean if I was to say to

huge area I mean if I was to say to someone think about a job I definitely

someone think about a job I definitely say integrate questions around ethics

say integrate questions around ethics and morals into this uh AI right now is

and morals into this uh AI right now is a hot topic how do we think about that

a hot topic how do we think about that but also more importantly again what are

but also more importantly again what are we solving for it's for the human being

we solving for it's for the human being to ensure relevance whatever that might

Um, just on the topic of of ethics and um and your comment about legal advisory

um and your comment about legal advisory potentially dropping. If legal advisory

potentially dropping. If legal advisory were to drop, could one infer a need for

were to drop, could one infer a need for ethical advisory to fill the gap?

ethical advisory to fill the gap? So AI applying rules, there will need to

So AI applying rules, there will need to be a human element to consider the

be a human element to consider the ethical implications.

ethical implications. Absolutely. Absolutely. Without a doubt.

Absolutely. Absolutely. Without a doubt. So the answer is yes. I was having a

So the answer is yes. I was having a conversation with a lawyer a couple of

conversation with a lawyer a couple of nights ago. She said something

nights ago. She said something interesting. I can't remember the exact

interesting. I can't remember the exact terms, but essentially what it is is

terms, but essentially what it is is this. There's there's there's the

this. There's there's there's the context is really important, right? So,

context is really important, right? So, what is the context of? So, if a

what is the context of? So, if a homeless if a if a if a poor mother has

homeless if a if a if a poor mother has a child uh and she she steals a loaf of

a child uh and she she steals a loaf of bread or thing to feed the kid, that is

bread or thing to feed the kid, that is a crime. It's black and white, right?

a crime. It's black and white, right? So, yes, that person's guilty, but

So, yes, that person's guilty, but there's a context piece over here that

there's a context piece over here that the machine won't be able to get.

the machine won't be able to get. There's a nuance behind them. There's a

There's a nuance behind them. There's a context piece. So it's not only the the

context piece. So it's not only the the it's not only the human element for

it's not only the human element for around the ethical implications

around the ethical implications downstream but also how do we apply that

downstream but also how do we apply that law in a certain context right it's

law in a certain context right it's about understanding

about understanding where do we bring in a level of and

where do we bring in a level of and within the legal precedence we have this

within the legal precedence we have this as well so I'll give an example of which

as well so I'll give an example of which in the western world by and large the

in the western world by and large the western world tends to tends to value a

western world tends to tends to value a younger life over an older life by and

younger life over an older life by and large but in the east it's flipped so

large but in the east it's flipped so the rule of law can't be the same across

the rule of law can't be the same across the board yes the law can tell us what

the board yes the law can tell us what is the baseline but we need

is the baseline but we need interpretation exactly the question that

interpretation exactly the question that you asked right now Jen there is a human

you asked right now Jen there is a human element to say what's the implication of

element to say what's the implication of that and while this might be the case

that and while this might be the case should we defer should we change this

should we defer should we change this somewhere else I'm a bit of a geek and

somewhere else I'm a bit of a geek and for those engineers that are that are

for those engineers that are that are that are listening right now hopefully

that are listening right now hopefully you'll enjoy this but I'm a bit of a

you'll enjoy this but I'm a bit of a geek and so I was rewatching Star Trek

geek and so I was rewatching Star Trek the next generation that's the one of

the next generation that's the one of Jean Lard and uh and data the machine

Jean Lard and uh and data the machine right and in the series data has his

right and in the series data has his name indicates has all the data but he

name indicates has all the data but he doesn't understand the nuance and he

doesn't understand the nuance and he doesn't understand the human capability.

doesn't understand the human capability. He doesn't understand the human stuff

He doesn't understand the human stuff and this beautiful interplay between the

and this beautiful interplay between the human gut sense around empathy around

human gut sense around empathy around warmth around humor and you got this

warmth around humor and you got this cold data statistical machine. It's that

cold data statistical machine. It's that interplay over there and that human play

interplay over there and that human play to what you said right now I think is

to what you said right now I think is integral moving forward.

J are there any more questions or do we need to

need to um No. So that's I think we do need to

um No. So that's I think we do need to wrap up. If there are any that come

wrap up. If there are any that come through, we can address those after the

through, we can address those after the fact. Um but

fact. Um but yeah, thank you so much.

yeah, thank you so much. Yeah, and from my side, thanks very

Yeah, and from my side, thanks very much, Jo. Thanks for having me and Jenna

much, Jo. Thanks for having me and Jenna and Silveroft. Um you know, folks, feel

and Silveroft. Um you know, folks, feel free to connect with me, carry on the

free to connect with me, carry on the conversation. And a big big thanks to

conversation. And a big big thanks to Jacqu to you and Jenna and Silveroft. Uh

Jacqu to you and Jenna and Silveroft. Uh I think we need more of these forums.

I think we need more of these forums. Again, please, as I said before, this is

Again, please, as I said before, this is just my view. I think there's multiple

just my view. I think there's multiple views out there. How do we get more

views out there. How do we get more people speaking about this? I think is

people speaking about this? I think is the key. I don't think knowledge should

the key. I don't think knowledge should be centralized and certainly there's not

be centralized and certainly there's not one person that knows everything.

one person that knows everything. Fantastic. Thank you Craig. Um I took a

Fantastic. Thank you Craig. Um I took a lot of notes and we look forward to

lot of notes and we look forward to engaging with you with you further.

engaging with you with you further. Hopefully I have another chance to speak

Hopefully I have another chance to speak together and uh and we'll speak soon and

together and uh and we'll speak soon and thanks to everyone for joining today.

Click on any text or timestamp to jump to that moment in the video

Most transcripts ready in under 5 seconds

One-Click Copy125+ LanguagesSearch ContentJump to Timestamps

Paste YouTube URL

Enter any YouTube video link to get the full transcript

Most transcripts ready in under 5 seconds

Get Our Chrome Extension

Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.

Add to Chrome — Free

Works with YouTube, Coursera, Udemy and more educational platforms

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube

←

→

↻

https://www.youtube.com/watch?v=UF8uR6Z6KLc

YoutubeToText

←

→

↻

https://youtubetotext.net/watch?v=UF8uR6Z6KLc

YouTube Transcript:AI & The Future of Professional Services Firms | Webinar Replay

Video Transcript

Paste YouTube URL

Transcript Extraction Form

Get Our Chrome Extension

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube Transcript:
AI & The Future of Professional Services Firms | Webinar Replay