How much could comms providers increase their profits if they made better use of analytics? Is there a conflict between the apparent certainty of mining data and the inherent uncertainty of managing risk? These questions are addressed by Mike Willett, a Partner in EY’s data analytics and information management practice. Now based in Auckland, Mike has broad experience of working as a manager and consultant for telcos in more than 40 different countries, including 6 years as Director of Fraud and Revenue Assurance at Australian operator Telstra.

Topical news items are also debated by the show’s three regular presenters, industry analyst Ed Finegold, senior risk executive Lee Scargall, and the Editor of Commsrisk, Eric Priezkalns.

Transcript (auto-generated)

Hello, this is The Communications Risk Show and I'm Eric Priezkalns. The last couple of
months we've been talking on Wednesdays about the risks faced by comps providers on their
customers. We're nearing the end of the second season. Next week's show will be the final
episode in the current run. But the point of the show always remains the same, to instigate
conversations about risk so we can learn what they are, what those risks are, and how to
deal with them more effectively in future. Now you can join the conversation at any time
during today's show by typing a message into the window immediately beneath the live stream
at tv.commsrisk.commsrisk.com. Your messages are anonymous, so add your name to the message if you want
me to read it out. And the show is also streamed live on LinkedIn and a bunch of other places
too. A member of the team will forward any comments left over at LinkedIn as we go along
and I'll try to read out as many of your questions and observations as time permits. Now later
in the show we have an interview with Mike Willett about analytics profits and the relationship
between analytics and risk. Mike is a partner at EY's data analytics practice. He's helped
clients in over 40 countries around Asia-Pac and Europe, and he spent over six years at
Telstra where he's director of fraud and revenue assurance. But first, let's catch up on recent
news and debate its significance with my co-presenters Ed Finegold and Lee Scargall. Ed is an industry
analyst, widely published author, and a strategic advisor to tech and telecoms businesses. He
joins us from Chicago. Lee is an executive and consultant who's managed risk on behalf
of comms providers in the Middle East, Europe, Caribbean, and Asia. He joins us today from
Bahrain. So guys, what the people who watch this show don't appreciate is that we've been
having this barnstorming conversation even before the shows begin, but now we have to
get back on script and focus on today's topics. So we're going to stop talking about all the
exciting things we were talking about just a second ago that maybe we'll talk about later
in the show. And today I want to begin by focusing on scam calls, scam texts. I see
a lot of social media activity about this recently because the US Senate held a hearing.
Oh my God, what a complete waste of two hours of your life that would have been if you'd
listened to that hearing. I can assure you there's probably not many sane people did
waste their time listening to it, only insane people like me. Scam calls, scam texts seems
to be a problem that never goes away. If you listen to some people, whack them all,
whack them all, never goes away, never gets any better, except in Australia, because in
Australia they're really good at reducing scam calls and scam texts. So at least the
number of complaints about scam calls and scam texts has been brought down dramatically.
So we're going to show a graphic on the screen now. There were 14,107 complaints about scam
calls in the 2020-21 financial year of the Australian Communications Media Authority.
Now they have a year end that ends at the end of June. That figure halved to 7,052 complaints
about scam calls in the 21-22 financial year. And then it fell again to 3,998 for the most
recent financial year. Very dramatic fall, a 72% drop in complaints about scam calls
over the course of two years. Contrast that with complaints about scam SMS messages,
which were 1,393 in the year 2020-21, then rose by 132% to 3,245 in 21-22, but has then
plummeted back down again, an 86% fall in the last year to just 465 complaints. That's 465
complaints compared to 3,245 the year before. The likeliest explanation for these drops in
complaints, the adoption of an aggressive call blocking program for robocalls, which became
mandatory in December 2020. And then similar steps being introduced to deal with SMS scam
messages, but they were made mandatory in July 2022, which would explain why the numbers went up
and then went down again. Their focus there with SMS messages is that commerce providers should
check that anyone sending SMS messages in bulk has verified the sender is entitled to use the
sender ID associated with that message. And they have been willing to find businesses that do not
comply with those rules. So I think this is very suggestive that their methods are working very
well indeed. So to begin today's conversation, Ed, regulators, as I say, have gotten into a very
bad habit of saying scam calls and messages are difficult to reduce because chasing the
scammers is like playing whack-a-mole. As I said, the U.S. Senate, if you'd wasted your time
listening to that, it would have been two hours of people going, oh, it's so difficult to catch them.
They always, oh, we think we know how to catch them, but then they go and do something different.
And we weren't expecting that. So then we have to do some, oh, pathetic. I've never heard anything
so pathetic in my life. Half a billion dollars, four years after they've last been in front of
the U.S. Senate and we've got nothing new to say. No explanation of how half a billion dollars has
gone up the wazoo, achieving nothing. My question to you, apologies, Ed, picking on you because you
are an American. Do these figures from the Australian regulators suggest that there are
some regulators who are making far too many excuses for failure? You always pick on me for
being the American. I'm used to it. That's just how the show works. That's cool. Absolutely, Eric.
You know, here's the funny thing is that when you first shared the story,
my eye was drawn to the chart and my immediate reaction, having been a publisher in communications
direction, what have you, was people, please, when you publish a chart, don't leave all the
label point sizes set to the default, blow them up so that when it gets published as an inset
and so on, like Eric's article, you can see all the stats on it. So that was my first reaction.
My second reaction was because I'm so jaded on this topic was, oh, complaints went down. Well,
what does that mean? So then I bothered to read your article and to follow the links.
And as I went, my confidence built. I was like, you know, I think there's something to this.
And I went ahead and followed all the way down. And if you follow the links, you'll find your
way to the action plan. They have a summary action plan that they actually published.
And on like page seven of that action plan, you can see the plan. And if you take the plan
and you take the results, you have a case study. And I think one of the things that
ought to get communicated out of this. And so here we go is this is a model for how to do the
scam call and text blocking. It's a model that works. They've proven it works.
So it needs to be widely communicated. And I think in that sense, there becomes no excuse
for not having some action to use this as a bar, in other words, that you can measure others
against. And there isn't an excuse to say that this can't work because we have a model that
does work. Now, I bet if you went to the Australian regulators and said, is it perfect? They'd say,
no, it's not perfect yet. We're going to keep fine tuning this. Right. But it's clearly effective
and it's a great model to start with. And so, again, I think that I went from being skeptical
to saying, oh, wow, hey, this is real. And they even told us how they did it. And so anything
we can do to communicate that I think is helpful to everybody. But you are absolutely right that
everybody wants to say it's whack-a-mole and they don't really want to do anything.
Well, now you don't have an excuse. And for me, I'm going to be controversial here because nobody
likes to talk about money because there's not a telecoms operator in the world who benefits by
saying we kept the amount of spending on consumer protection down because you put that out into the
public domain and people are going to get upset because they're going to just go typical business,
cutting the corners and being cheap. OK. But for me, I like cheap, cheap that works,
because cheap that works is a hundred times better than expensive that doesn't work.
You get it done. No one's got an excuse not to do it because everyone can afford to do it. And then
you get the results and you move forward. Whereas the U.S. approach for me is the epitome of
expensive and doesn't work. Lots and lots of money spent. And then four years later, you're
going, yeah, well, you know what we need to do? We need to spend some more money. Oh, OK. I mean,
don't bother to learn a lesson from what was previously spent. And so the contrast for me
is extraordinary. If you look at what the Australians are doing, it's cheap. It's simple.
It can be replicated. But that's why no one's out there banging the drum because no one's really
selling this. No one's got a great incentive to go out and say, you must do the same here
because there's no salesman, there's no marketer, there's no profit to be generated. It's just good
common sense consumer protection. Anyway, that's my view. Lee, let's bring you into the conversation
here. Now, we've often spoken on this program about some of the mitigations, common sense
mitigations for nuisance calls being adopted in the Middle East because other parts of the world
we talk about because other parts of the world just are not aware of what's happening in, say,
the Middle East. We've also had representatives on the show from the Brazilian regulator Anatel
earlier this season discussing their approach because, again, people didn't know the approach
they were taking. Why, Lee, my question to you, why is the global electronics communications
industry, and note how I describe them, why is the global electronics communications industry
so poor at telling people around the world about when they're successfully reducing scam
calls and messages? I do feel sorry for the Americans, Eric. You're always picking on them
week in, week out. Look, you're not going to repeat what you did last week. You're not going
to ambush me with that pro-China speech. Don't play devil's advocate with me. I've been bigging
you up here. I've said you did good work in the Middle East. Stick to the script. Don't be just
jumping on that bandwagon of getting kudos and favours by opposing me. You know I'm right, Lee.
Why are we not talking about success stories like the success stories you've had in Bahrain?
You are right, Eric, and I do agree with you. This is a plague which actually affects the entire
world, right? So I've got a couple of stats here. Spam calling in Indonesia and Vietnam,
that has actually doubled in the last 12 months. In Indonesia, they actually receive 14
spam calls per person every month, right? And they have a population of 280 million people.
In the US, you mentioned the US. I was looking at a statistic about them the other day.
It said that 23% of Americans, they actually reported they lost money as a result of scam
calls in the last year, right? And that the average there was around 500 US dollars. So
if you actually look at the trend in scam calls, it doesn't look good. But there are signs of
successes. We just spoke about Australia there, which you mentioned they've got a nice cheap
solution, which is reliable, and it's looking like it has success. We had the guys on from Brazil
the other month. And also, as well, last week in Saudi, and we had a discussion about this the
other week, is that they actually launched their own version of the Caller Name Presentation or
CNAP. Although, you know, I've tried to find a bit more information about that on the internet,
but there doesn't seem to be much information. So I think there are good signs of progress out
there, but it's just been underreported in the news. But the good news is, if you want to hear
about these stories, then continue to watch the COMSRIS show. Well, it shouldn't be up to us,
should it? I don't have the resources of the FCC. Did you see what the FCC has got now? I'm going
to just continue the conversation. Now, we haven't prepped for this. But this is hot news. This is
like only yesterday was announced. Jessica Rosenworcel always manages to get her name on
every single press release that makes it look better. And her name is now attached to a press
release where the FCC is going to look at using artificial intelligence to stop scam calls and
scam texts. I'm looking at a situation where Australia has had tremendous success, and we're
not even using fancy words like artificial intelligence. Ed, is this another example of
basically the FCC stuck in a rut where they will look at something if the solution involves spending
money on technology, as opposed to learning from other companies' examples, other countries'
examples? Yeah. I mean, again, I feel like there's constantly these examples of just the crowd in
Washington being completely disconnected from what's happening on the ground and rather being
influenced by what the buzz is, what the hedge fund crowd is telling them to talk about to pump
up. So everybody's talking about pumping up AI right now. It wouldn't surprise me in the least.
And again, keep in mind, this doesn't only happen to the telecom sector. I seem to recall this
happening in solar panels sometime in the recent past as well, where that then became the hot topic
and some company got a load of money from the federal government and it turned out to be a scam
in solar panels. And so it wouldn't surprise me to see the same type of cycle here
around AI when, again, when you contrast that to page seven of the action plan.
Right? Like, come on, you want a solution? Go to page seven of that action plan. If you're
talking about AI on this, that's where you're starting, then yeah, you're starting that same
hype cycle. And then you're feeding, somebody's telling you to feed that, right? To enrich
themselves. And the outcome is so predictable because in four years' time, it will be,
oh, we could never have predicted that the fraudsters would use AI and the fraudsters
will be better at masking and hiding and disguising their traffic than the ordinary
people who are going to get picked on wrongly by the false positives of AI. There'll be yet
another investigation and there'll be more technology thrown out to solve the problem
instead. Is that not the case, Lee? Am I exaggerating? Do you share my fear, Lee,
that AI could be very dangerous indeed handing over the decision about which calls,
which text messages are blocked to effectively computers when no one's clear about how they work?
We're using AI at the moment, or we're using technology which is kind of branded AI,
but I'm pretty certain the methods of using it has been around for years, right? I mean,
it just, it's given that, everything's given that name AI. I think it's just more of a sales term,
but yeah, there's some kind of core characteristics that you look for and then, you know,
it's fraud, right? But I agree with you. I think if you look at the US example, they do like to
chuck technology and money to solve a problem when there's a lot simpler solutions out there.
Well, thanks guys for agreeing with me. That must be a first there. Both Ed and Lee agreed with me,
despite Lee's temptation to ambush me and gang up against me and be all pro-American.
FCC, if you're listening, pay attention. Comms risk is where you hear what's going on around
the world, not those stupid committees that you're getting in Washington. But now, an ad break,
an ad break, because we too believe in free commerce after all. It's time for our interesting
Fraud Fact of the Week, courtesy of the PRISM Fraud Intelligence Team at Symmetry Solutions.
Now, most people take phone numbers for granted, but the abuse of phone numbers has become so
severe that the US state of Maine has passed a new law to tackle it. They found that some
communications businesses were buying up huge volumes of phone numbers so they can be used to
evade anti-fraud and anti-spam controls. Bad actors work around analytics that attempt to
identify anomalous call patterns by using a different originating number for every single
call they make. Buying lots of phone numbers is not a crime, but it could be indicative of crime.
Now, the solution adopted by Maine is to impose an additional levy on businesses which purchase
numbers using the state's 207 area code. This will not stop abuse activity, but it will make
it a lot more expensive for the abusers and the businesses which facilitate abuse. So, to learn
more about PRISM's Fraud Intelligence Services and how you can be learning about numbering
intelligence to defeat fraud, contact Symmetry Solutions. Their URL is symmetrysolutions.co.uk.
Now, back to the chat, guys. So many things to talk about. It's so difficult to fit it all in,
and I think this one, I really don't know how much time I'm going to need to spend on this,
but it's such an involved topic. The European Commission used micro-targeted adverts on apps,
formerly known as Twitter, designed to get support for a change to the law to require
internet platforms to scan messages for words and images indicative of child pornography.
Now, whether that's a good idea is a debate in itself, but let's focus on what exactly
the team of European Commissioner Ylva Johansson, who has the home affairs brief in the Commission.
Numerous objections have been raised to her proposed child sexual abuse regulation,
with opponents labelling it chat control because it will create a universal mechanism to control
what is said in emails and direct messages. Some of those objections have come from members of the
European Parliament, others from participants in the European Council, which represents the
national governments at the EU level. Now, for those unfamiliar, the EU has a three-way mechanism
for setting laws. European Commission is the executive body, but all laws need to be agreed
by both the directly elected Parliament and by the Council, which represents the national governments.
Leaked notes from a September 14th meeting of European Council found that many countries raised
objections to Johansson's regulation. The next day, Johansson's team ran adverts on apps in those
countries, the adverts designed to encourage the public to support the regulation with scary images
of children and adults who may be interpreted to be predators. And they gave the message that
time is running out, creating that sense of fear, the need for immediate action.
So please, producer James, please share the adverts with the audience now.
So, very emotionally charged material there. A short advert, but used widely across the European Union,
let's say targeting the countries where the governments, the national governments, had raised
objections or had raised reservations about the new regulation. Technology expert Dani Mikic
reviewed the transparency reports of X and he found not just that the adverts are running in
the countries that raised objections, Belgium, the Czech Republic, Finland, the Netherlands,
Portugal, Slovenia and Sweden. But here's the issue I really want to focus on, guys.
This advert was targeted at people who were perceived to be more likely to be supportive
of this new regulation. That was done by choosing to exclude users of X who had shown an interest in
any of the following topics. Christianity, Julian Assange, Brexit and its Dutch and Spanish equivalents
Nexit and Spanexit, Viktor Orban, Nigel Farage and the Alternative für Deutschland, a German
right-wing political party. So, to put it simply, groups who may be sceptical about the EU's powers
already, especially sceptical about surveillance powers in the EU, people who were interested in
those groups were not shown these adverts. So, my question for my two fellow presenters.
Some argue that micro-targeting adverts in the EU based on political and religious beliefs
is already illegal in the European Union because the European Union only permits the processing of
sensitive types of data, including such beliefs, for strictly limited reasons. Whether this was
legal or not, do we as panellists believe what Johansson did in this instance should be prohibited?
Let's start with you, Lee. So, that list you just put up there, actually, that's a subset of the
full exclusion list and I actually went through that list and the two things which stuck out for
me was one was Qatar and Catalonia were on that list. So, I'm not really sure what, you know, in
particular the Qataris have done to upset the Europeans there, but anybody who had an interest
in Qataris or Qatar wouldn't be shown that particular message. But there's a bigger story
here, is that the author, this Dani Mekic, he was actually de-platformed from Twitter or X,
right, and they didn't even provide him with an explanation as to why they did that. They gave him
the whole Russell Brand cancel culture to him. But look, first up, it actually breaches X's
advertising policy, which actually prohibits religious and political micro-targeting.
And second, religious and political views are actually classed as sensitive data, so it actually
breaches the EU's own Digital Service Act. And I believe, actually, and I had to laugh at this,
is that the EU have actually opened up an investigation into themselves regarding this.
But look, the whole thing actually reminds me of Cambridge Analytica all over again,
right, where they're targeting specific groups of people who are likely to vote for a particular
policy or a person. But do you think it should be prohibited? Do you agree with the way the
rules that you've just described should work in practice? It should be banned.
Okay, that's clear. Ed, what's your feelings on this? Do you agree? I mean, bearing in mind that
the European Commission here, the Home Affairs Brief, this is the person who's responsible
for rules like this, is saying that they've done nothing wrong and that they've done nothing
illegal. Whether it's legal or not, do you believe, do you agree, that activities like this,
this kind of micro-targeted advertising based on someone's existing beliefs, do you believe it
should be banned? Yeah, I mean, so I agree with Lee that especially this specific type of case
should be banned. But rather than just repeating what Lee said, I want to maybe spin the whole
thing around and relate it to another issue we've discussed, which is everything that relates to
stir, shaken, CLI, like all the caller IDs spoofing, all those kinds of things and what
the intent or philosophy behind it is. And so we've talked a few times on the show about how a lot of
times we feel like the regulatory regime is trying to figure out how to protect telemarketers,
right? And how to sort the bad from the bad spam calls from good spam calls, right? And it's been
from that perspective. And so I want to bring that similar mentality here to what we talked
to the Brazilian regulators about, which was like, hey, a person should have the ability to decide
if they want to answer a call or not because they're given the proper information.
Why can't the same be true of ads and micro targeting and personal personalization?
And we talked about AI a minute ago, and I find all I find all of this
disturbing. And if you want to put the disturbing cherry on top, it's that
having walked around, you know, like events and
Dean, all the type of advertising that Lee's talking about, right, the AI labeling on things,
which is like gluten free now on technology, right? It's that a lot of the discussion of
the application of that on the customer side, for example, is for more personalization.
So I say again, like, what are we trying to do here? This feels like a pros versus cons kinds
of thing, right? And so I come back to like, the reason a lot of this stuff should be banned,
and that we should probably take a much harder look at micro targeting in general is because
I don't think anyone's ever looked at micro targeting and really said, like, should we
be doing this? What is the purpose of this? Who does this benefit? Right? I'm sure Google's very
excited about it. But what about the rest of us? Right? And so that's, again, I just would spin the
whole perspective of it around and tell people like, hey, you, you do have a right to decide
whether or not you want to be advertised, you ought to, we don't want to give people the right
to decide, we want to tell people what's good for them. That's, that's the fundamental philosophical
debate that's occurring here. Except it's a, it's a very skewed debate, because people in power
want to do the talking, and they want the rest of society to listen. And those of us who don't like
the idea that we have to be sit, sit down, be good citizens, just listen to what we're told to,
and we're not allowed to listen to anything else. We don't get a say in the story. That's why the
big government agencies, the big government actors want to talk to the big corporations
and stitch up the deal between themselves. That's why the arguments at that level, it's not about
setting individuals free, or protecting them from an abuse like this. This seems to me,
Ylva Johansson has engaged in a very serious abuse on many levels. She's broken European law that
she's responsible for enforcing. She's done it in order to undermine the will of democratically
elected governments around Europe. She's done it by, by going out to specific individuals,
you know, using the power of the algorithm, to find specific individuals who are most likely
to agree with her, frankly, provocative. At the very least, one would say, emotionally charged
message. Time is running out. What exactly is this time that's running out? I'm not in favour
of child pornography. Many people who are against child pornography, though,
can still raise a question about whether it's wise, it's sensible to implement scanning,
automated scanning of every single electronic message. Because once we implement the scanning
for child pornography, we've got in place the mechanism to scan every message for everything
that we might want to find in a message, which will not be just limited to child pornography.
So there is another side to this argument, and the idea that we must be barrelled forward at
breakneck pace. I think this is a very worrying sign. And it's a sign that we have people in the
European Union who are very happy to override law, because they're always telling you,
we will protect you, we will keep you safe. Same people have been saying the same thing about data
protection, privacy, and all the rest for donkeys years. I don't feel any safer. As a result,
they've created an illusion. Lee, jump in here. I can see you want to talk.
Yeah, you just showed the Twitter ad there. And on that Twitter ad, there was some graphics,
some statistics, which they showed. Now, there's actually a lot of questions
surrounding the integrity of that survey and how it was conducted. The questions,
they've been through it, the questions, they only highlighted the benefits of the regulation,
didn't actually consider the consequences of mass surveillance. So this has actually led to the
biased statistics, which have been shown in that ad. So it's also, it's a misleading advertising.
And look how divisive the approach is as well. It's feeding information that people want to hear,
creating a split between people who are good people, and in the view of the European
Commissioner here, these people we like, these people we don't like. And I can see how the
division works. Viktor Orban. Well, a lot of people in Europe don't like Viktor Orban. A
lot of people think Hungary is on the wrong track. Okay. Alternative for Deutschland. Well,
again, obviously the majority of Germans vote for other parties, only a minority of Germans
vote for that party. Nigel Farage. He's an often demonized figure considered to be like,
next to the worst of devil. Let's not even let him have a bank account in the UK because he's
such a demonic character. We can't allow him to even live or work or do anything. And it goes on
and on. And then Julian Assange gets added to the mix. Okay. Julian Assange. I mean, I think there
were some people who thought that he had a good point to make about whether it was good or right
for American attack helicopters to be shooting civilians just because they were in the wrong
place at the wrong time. And then Christianity. How far do you go down the path of labeling some
people good and some people bad? As you said, Lee, Qataris, Catalonians. This is about dividing
populations. Microtargeting for me, the danger, the greatest danger with microtargeting is about
setting people against other people, sending selective messages to some people about how
they're right. They're good. They need to impose their will on the others because they can't be
trusted. And that's exactly what's happened here, isn't it? How many of those people, even the ones
who would say they don't like Victor Orban, they don't like alternative for Deutschland,
they don't like Brexit. How many of them are going to feel comfortable that now Christianity
is being presented as being somehow in the evil camp as opposed to the good camp? I think that's
a very dangerous and scary precedent from a politician who's effectively showing her world
view to the rest of us. She thinks Christianity is on the side of the evil. She thinks Christianity
is on the bad things. She thinks Christianity has got something to do with Brexit and that the two
have got something to do with being too relaxed about child pornography. I lose the words to
express myself here, guys. Am I, again, going too far? Am I too overboard with my emotions? Because
I think this has not gotten much coverage, this story, and yet this should be the biggest story
in the European Union right now. You compare this, say, to Cambridge Analytica, Lee. Cambridge
Analytica is a walk in the park in comparison to that. That was the political party engaged in a
campaign and the campaign techniques may or may not have been successful. This is the European Union
using its power to manipulate voters within Europe. Am I going too far? Am I exaggerating?
Well, it's breaching its own laws as well, Eric. I mean, she needs to resign. She needs to go.
She's not going to go. Let's be realistic. She's not going to go.
No, because nobody's held account over there. That's why.
You know, this is a hardliner. This is a political extremist. This is an ex-communist.
She's had an entire career based upon being told everything that she wants to believe is true is
true, and she's now in a position of power in the European Union. She's not in any way accountable,
and even the media is not holding her to account. Ed, is there a failure on the part of the media
to not properly report these stories? I mean, you know, in the U.S. we wouldn't get a lot of
coverage of this kind of thing to begin with, so it's a little bit hard for me to say. But I think
the bigger issue here that you're raising, I think it brings into light which would apply to
this person or any other person, right, in such a position of authority is effectively
what this person is doing is the same kinds of things that,
you know, the various negative state actors have been accused of doing, right, in their
intelligence agencies in terms of interfering, right? So, like, the Russians have been accused
or found to have done in terms of interfering in elections and other public policy, you know,
matters in foreign countries. And basically, you're doing the same thing, taking the same
kind of actions within, but within your own country. It's not any better. I mean, it's as
bad, maybe it's even worse, right? And so how far down the chain you go to say that, you know,
you're being a traitor, right? You know, that's not necessary here. I think it goes back to kind
of what Lee was saying, which was like, there's a lot, and I think what you said, if you break it
down, there's a lot of serious laws that are violated. And I think where this gets touchy is
the, boy, it's a little bit hard to tell the difference between legal micro-targeting and
illegal micro-targeting, right? Which I think maybe you had raised the question at some point
as to whether and how much of micro-targeting should be legal at all, because of the degree
they would be abused, which where you get into that like pro versus con discussion of what are
the pros versus the cons of allowing this kind of use of personal data for advertising and marketing
reasons. The cons we know are very heavy, disruption of society and democracy. I mean,
go down the list. There's a lot of heavy ones. Are there enough pros on the other side? Like,
you know, Google gets to make money. I mean, right. So that, I mean, that's why this whole
discussion is so bizarre to me, because none of the incentives seem to be in the right place.
And so then we end up seeing these really bizarre abuses, like you're calling out here by someone
in authority, right? And, you know, when you break it down and you see just how bizarre it
gets and what's actually, what I was thinking about as you talked about that, that maybe is
even stranger is someone had a meeting and went down that exclusion list and they didn't think
about any of the societal aspects that you thought about. What they thought about was how much money
am I going to spend on this ad campaign and who am I going to pay to see this? And so it was the
way to get inside the budget was to make the exclusion list bigger. And that's how Christians
ends up on the list. Right. And it's, and, but that's it. There's no other thought that goes
into it. Right. Like you're in a different lane of reality where all you're thinking about is
how am I triggering the algorithm to accomplish my objective, as opposed to having any thought
of like, should I be doing this? And is this moral? And, you know, no, it's not even in the
equation. I think at that point, can I, can I just add here, you know, whether this is legal or not
and whether micro-targeting is good or it's bad, we're not talking about the real issue.
The real issue here is mass surveillance, right? Now we should be talking about that,
but that's probably going to be, we could do a whole show on that.
There's so many layers to this conversation. I mean, forgive me to, you know, to reiterate the
point that you've both been making for me. For me, this is about surveillance. Yes. And how do
you persuade people that they should want surveillance? You use the techniques that
when you are being disparaging about the techniques, you brand it populism. Nigel Farage
branded a populist, Viktor Orban branded a populist, Alternative for Deutschland
branded populists. They get branded as populists when you don't like it. And then you use the exact
same techniques of dividing society, one group, another against another group and fear.
We must act now. There's no time to waste. You must allow us to read your messages on all the
social media platforms, all the emails that you send. You must allow us because otherwise there's
too many people going to suffer. There's too great a risk. It's an emotionally charged message. They
know full well what they're doing with that emotionally charged message. And it's exactly
the same kind of technique that they then complain about people on the other side of the fence also
using emotionally charged messages. It can't be right to say that emotionally charged messages
can't be used by one side of the debate, but we're going to use it on the other side of the debate.
I think the European Commission here has really been revealed as being a great cesspit of hypocrisy
on this particular issue. They like to pretend that they're the arbiters of good behaviour,
of good common sense, centrism between Europe. A lot of people have a very romantic notion about
how the European Union is run. And yet we've got a European Commissioner in a very important
position of power, very clearly abusing her power. The pro-euro crowd, they almost don't
want to talk about it. They don't want to deal with it. And I don't think it should be up to
supporters of Viktor Orban or Alternative for Deutschland or Nigel Farage or Julian Assange
to point out this is wrong. This is for everybody to point out this is wrong. Whether you're in
favour of increased restrictions to protect children from pornography or not, you should
still be pointing out that this kind of abuse by somebody in power must stop for the health of
democracy within Europe, within European countries. And so, yeah, I'm very scared about being bumped
into yet more. We already have too much of this. Using scare tactics to get even more surveillance
will never be able to roll it back again if these people succeed.
Put my head back in that room. Yeah, no, I'm because as you're talking, I'm putting my head
back in the room and thinking about, you know, someone that's executing a campaign like this,
right? And what you're talking about, like the degree of fear mongering, which I'm tired of,
right? I'm tired of everybody pushes the fear button. It's like, yes, thank you very much.
There's whatever 20 books everybody's reading about how you do online marketing that says fear
is the number one thing, you know, that you push that button. So everybody pushes that button
because fear sells, right? And it's annoying at this point. And you could see like the meeting
unfolded where it's you're sitting down in front of whoever makes the decision for the commissioner
that they're going to run this ad campaign. And they're using taxpayer dollars to pay some
consultant to come in and tell them, right, here's what your exclusion list should look like,
including Victor Orban, and Qatar, apparently. And right, here's right. And here's the top three
emotions that you could use. And we recommend fear, because that proves to be by 47%,
the most effective, right, that you could see that unfolding. And that's how these things get
normalized and neutralized. And they become very separated from the reality of what it actually
does. Because now it's just a discussion about algorithms, and numbers and budget. And we're
done. Right? And that's it. Yeah. And it's, that is disturbing, though, right? Like it disturbs me,
because it makes it very easy to make dehumanizing decisions that way.
Yeah. And what did we talk about earlier in the show? AI for deciding which calls and
texts are going to be blocked. It's algorithms again, and again, and again. And we're supposed
to just trust that people in power will use them well and use them right and use them responsibly.
I mean, it's whether you're spying on a message, or whether it's deciding who gets which message,
or blocking communications from a person to person. We see again, and again, and again,
these topics coming up. And the quality of the public debate, the quality of debate in the media
is so low. It's way too low. No, I mean, every time you bring up the threat of artificial
intelligence, some WALL-E will go on about, well, they're not going to build Terminators,
you know, Terminators. Yeah, I know they're not building Terminator robots. Okay? That's not the
risk that we're worried about here. Okay? Is this the quality of debate we have, you know?
If that's the quality of debate you have, of course, we're going to end up with a terrible
number of abusive situations where artificial intelligence and other algorithms is going to
cause terrible damage to many people. Because if the only thing you're worried about is whether
some Arnold Schwarzenegger clone is walking down your street with a great big machine gun,
you didn't understand the problem, you know? And they don't want you to understand the problem.
They just want you to hand over the power, trust them, they know what's best, they've got to be in
charge. And I'm getting scared. I'm getting scared. I think we need to move on to an ad break,
because I need something happy and positive to talk about. And I need to listen to somebody
with a happy, positive, optimistic tone. And it's not going to be me, because I'm just stuck in like,
scared mode now, because the European Union, yeah, drive me crazy. So here's another one of our
regular sponsored features. And this is from our friends at OneRoot, the experts in co-authentication,
fraud prevention, and geolocation. Now each week we travel around the world via the phones
in our pockets. We see the good side of the world via the phones in our pockets. And this week,
OneRoot's Alyssa Giles is going to take us on a trip to Aruba. Producer James, please roll VT.
Hi everybody from OneRoot, I'm Alyssa Giles, and this is The World in Your Phone. Let's talk about
Aruba. This country is a flat, riverless island renowned for its white sand beaches.
Its tropical climate is almost constant at about 27 degrees Celsius or 81 degrees Fahrenheit. Did
you know that in September 2022, Aruban State-owned telco, Satar, announced its focus on a project to
install fiber broadband access technology in the north district of the island. The company promised
reliable, stable internet with super speed. They also added that the operator will continue to
invest in a new and modern infrastructure for the island, with other districts also earmarked for
fiber to the home installations in due course. A few of the other interesting facts that I found
about Aruba include that Aruba is not in Hurricane Alley, which means there's no bad time to visit.
Since the crime rate in Aruba is pretty low, it makes it one of the safest islands in the Caribbean.
Something that's wildly popular in Aruba is beach tennis. Think of a mix of tennis and volleyball
and played on the beach. Another fun fact is that you could fit Aruba into Ireland 471 times.
Iguana meat is a local delicacy and apparently it tastes like chicken.
Aruba is home to wild donkeys and there's a donkey sanctuary for injured and mistreated donkeys.
Be sure to subscribe to OneRoute on YouTube where you can catch up on the world in your phone
and watch OneRoute Roundup, the show that spotlights individuals and companies making
a positive difference in the telecom industry. I'll leave you with one last fun fact about Aruba.
The only poisonous creature on the entire island is the rattlesnake. Now Eric, back to you and more
of your awesome communications risk show.
Thanks Alyssa. I'm going to come back to the debate that we were having before because
it's so loaded. We literally cannot stop this debate and apologies to everyone who's waiting
for the MyQuillat interview, which is a really good interview by the way, but we're going to
just continue the debate a bit longer here now because this is such a thorny topic. Ed was just
saying here he's very upset that billionaires are going to drive us all crazy and I'm saying,
that's not the billionaire. I'm not worried about the billionaires. I'm worried about the people in
government who are going to do it. Lee, where are you? Would you pick on the billionaires or would
you pick on the people in government? Who scares you more, the billionaires or the people in
government? Both. Typical fence sitting answer from Lee there. It's a continuum. But you raised
a good point about artificial intelligence, the latest scare story about artificial intelligence
that you wanted to bring up. I did, and it was an article I read today. I think they're trying
to demonize AI because in this article, some people have discovered there's about 3,000 images
online of child pornography, which had been generated. They have been generated by AI.
So in the press, it's all anti-AI. Don't get me wrong, child pornography is the lowest of the
lowest, so bad. But the way they're going about it with this mass surveillance, got to stop AI,
all this type of stuff, the media is really trying to push it into trying to get a certain narrative.
I don't think the press cares either way. I disagree with you. I think they just vacillate
from one extreme to another extreme to generate whatever number of clicks, views, newspaper sales
they can. They'll go from AI is paradise to AI is doom to back again and back again. And this is why
when you look at surveys saying that young people are distressed, they feel a lot of anxiety,
they see no purpose in life. We are flooded with doom and gloom messages, then backwards and
forwards like a table tennis ball, like a ping pong ball between it's all going to be great,
tech is going to be this, and it's all going to be a disaster. We have no happy medium
for discussion anymore. We have no common sense. We have no rational centre ground anymore. Even
people who claim to be in the centre ground are stepping up and saying, look, some of this was
utterly predictable. How stupid do you have to be to not understand that if you can use artificial
intelligence to generate an image from real images, then it's going to be abused by pornographers.
Of course it's going to be abused by pornographers. People want to see dirty pictures. Here's a new
way to create pictures. Why are we surprised it's being used for dirty pictures? It's the most
predictable thing possible. You know, Eric, tech is never, that's what I was saying to you earlier,
tech is never sold that way. And AI is something that has been heavily bolstered
by the tech hype cycle. And this is where the tech hype cycle actually can become dangerous.
Heavily, heavily bolstered by the tech hype cycle. And that starts top down and it's who
are the big money investors and what are they pushing out into the market at the big conferences
that you know about and who's swallowing those lines, hook, line and sinker and bringing out to
the market. And the media can feed off of that because the cycle goes on and on. Now it's AI,
then it'll be something else, but they can just keep telling the story and they build it up and
tear it down. And I think part of what happens here with what's happening with AI right now is
it's like one of the oldest parables in Western culture. It's like the billionaire is the high
priest and AI is the golden calf and he holds it up and all the natives get down on their knees
and want to worship it. And it's like, guys, are we not going to learn this about the technology
is not going to save you, right? This idol that you're worshiping is not going to come in and
save you. And the billionaire high priest who's selling it to you is not really interested in
anything other than getting the money from you. Let's understand that there's a direct connection
between those two things. You're using the word billionaire too much and I have to read out a
comment from Bartok here to put you back in your place because there's too much anti-billionaire
stuff. I know Bartok, thank you for watching Bartok. I know you're more of a fan of billionaires
than Ed is. So I'll read out a comment here. Building a European, I'm just going to read it
out. I don't care. Building a European socialist slash communist super state is underway. See,
there you go. Now, any opposition is being branded populists, lose jobs, your bank accounts,
but it continues this direction. Soon there will be concentration camps, gulags in the forthright
or union of socialist European republics. This is why I love Bartok. He doesn't hold back with
his opinions. Some of us could see this coming a long time ago. Stalin, Mao lovers already headed
up the EU. Romano Prodi, Jose Manuel Barroso. Even this message may be in future be used to
sort me out. In fact, we're going to be cancelled in future, Ed. Not by the billionaires, but by
big government, okay? So Bartok was a bit worried about putting his name in it, but I've read out
his name like several times. It's too late now, Bartok. I only read that bit about your comment
right at the end. So Bartok is in trouble. We're all in trouble. It's not the billionaires, Ed.
It's the governments. The governments are ultimately got more power than the billionaires.
Yeah, I come from a country where those two things are very closely connected on a legal level,
though. Keep in mind. I see that as a continuum. That is two different-
How much does the US government spend every year, Ed? There's no way that billionaires
are spending anything like what the US government spends. Half a billion dollars on stir shaken.
Where did that get anybody, I tell you? Who has the most influence though, right? So,
I mean, we don't have to get into this debate on our show for the sake of our audience, but
you know, who in the end has- We'll do a new show in future.
US Congress makes, right? We'll do a new show in future for our
politically inclined viewers on Rumble, where all the bad people, where all the Russell Brands and
Glenn Greenwalds of the world are on. We'll go on there. We'll do a show like that on this one,
if you're both up for it. Bartok will be watching us there and we won't get any sponsorship because
actually we don't want to be too mean to billionaires, Ed, because I wouldn't mind
having a few more sponsors on this show. So, we do like private enterprise despite what Ed says.
Now, let's move on. Let's move on because we've not even got around to the interview that we've
got recorded for today. It's actually a really good interview, so I feel bad now that we've
overrun so much, but it's been a great debate, so it had to be done. So, right. Today's interview.
It's a pre-recorded segment, an interview with Mike Willett about making the most of analytics.
It would have been great to have Mike live and I would have loved to have his views on this too,
but he does live in New Zealand, which does make it a bit tricky to get him live. It's probably
about 5am, I think, in the morning, right there, down there now. So, appearing live,
not feasible for Mike. He's a partner at EY's data analytics practice for the Asia-Pacific
region. Previously worked as a consultant based in the UK and then spent six years at Telstra
in Australia, where he was director of fraud and revenue assurance. In this interview, Mike talks
about how to increase profitability by making good use of data. See, we're not totally against
using data or making money. I'm in favour of making money and data responsibly. So, Mike is
going to talk about this in this pre-recorded interview and also the relationship between
analytics and profitability maximization and risk management. Well, enough of me, enough of us.
Producer James, let's roll the VT. Hi, Mike. Thanks for joining us today from New Zealand,
where it's still early in the morning for you. Now, since 2017, you've been a partner at EY DNA,
EY's Asia-Pacific data analytics and information management practice,
but your background was in telcos. You joined EY after spending six years at Australian telco,
Telstra, where you were the director for fraud and revenue assurance. Now, first thing I wanted
to ask you about today is to draw upon your experience of telecoms and relate it to what
you've been learning about data and exploiting the value of data since you've moved on from Telstra.
Clive Humby, he coined the phrase, data is the new oil, back in 2006. Now, you will remember,
I'm sure, that telcos had a lot of data even back in 2006. They also have a lot of data now,
and that's pretty clear from anybody who follows the headlines about data breaches.
When you think about your career and what you know about the value of data and how companies
extract value from data, are telcos maximizing the value that they could extract from the data
they possess? Which kinds of businesses would you say are doing the best job of getting value
from data, and what should other businesses like telcos learn from their example?
Yeah, it's a great question, and hello, great to be talking to you again, Eric. Well, I think
there's always room for improvement for all organizations, whether they're telcos or not,
and we see lots of organizations, whether public or private sector, making significant
investments in exactly that question, how do I get more value from my data? Are telcos doing as
well as they could? I'm sure some are doing some great successes and some fantastic case studies or
user stories that they've delivered or use cases back to the business in terms of getting value
from it, but that's not necessarily universally spread across the whole of the organization,
and that's kind of what we see or I see in many clients today. There are pockets of greatness,
which could be really driven by individual champions or heroes who are doing amazing kind
of work, but it's not spread across the whole of the enterprise, and so that's probably the
opportunity for most organizations today to spread that across multiple parts of the business,
not just in places where it's particularly been strong historically as well. So we know,
for example, financial institutions are spending a lot of money and a lot of resource and effort
using data to meet their regulatory obligations. We know web companies are doing a lot around the
customer kind of agenda as well, and where I've spent a lot of time in the manufacturing business
looking at how they can improve shock front experience or the floor experience as well as
the supply chain areas. So I think you'll find organizations, they'll focus on what
their most strongest at and they'll get value from data there, but there are lots of other spaces
where they can do much more things. A fascinating answer. We could pick that apart so much because
I'm not sure if you told me if you thought telcos are or are not maximizing the value of the data
they possess. We know they've got a lot of data. Oh well, okay, then the answer would be probably not.
If you want to take up maximizing the value, I'd say no, there's definitely room for improvement
there. Where those improvements are though, perhaps, is that up to each individual telco,
depending on where their existing strengths may lie. So I can't give you a general view to say
telcos must do, all telcos are good here and all telcos are bad there, but I'm sure, as I said,
they're not maximizing as much value as they could today. Okay, that's intriguing. Intriguing, you
haven't pinned down to a generalization, but I love that answer. Now this show is about risk
and you use data to manage specific kinds of risks for telcos. Should businesses be thinking
more broadly about how to engender a data conscious culture, instead of just using data
to identify and tackle a narrow selection of risks? And I obviously ask that question because
you come from a background of managing some very specific risks on behalf of Telstra.
Yeah, well, I think the traditional approach to data was to use some analytics,
determine whether risks were evident or not, or whether the control environment was operating
as would be expected. But today, I think we're all aware of some of the new and emerging
technologies and GPT is no doubt the one on most people's minds at the moment, in terms of
its potential, but risk professionals now can extend kind of their remit to say,
how are these tools being used? Are we putting confidential information into the public domain
where these models are learning, essentially our secret source or the IP that we have?
So, I think the data focus now for many risk professionals is not just using data to identify
the risks, but how are others, and this comes to your maximizing question, but how are others
in the organization also protecting the data and the IP of an organization?
As well, and what are those uses? Particularly, because we are aware that everyone knows that
the tooling now is much more democratized. It's a lot easier for people who are perhaps less
experienced in data to get a subscription on the cloud and go and do some really interesting and
cool stuff at a lower barrier to entry. That also increases the level of risk with using that data
in the right way. Are you saying that technology has reached a point now where the technology for
exploiting data, for interpreting data, for drawing conclusions from data, has reached a point
where you don't need to be a technologist to actually maximize the value of data?
I think so. There's certainly a range of tooling out there with a little bit of smarts and a lot
of business context that people with a lower technical level or degree of competency
can extract some pretty meaningful insights from. You can't just ask a computer a question and get
an amazing answer, so it's not at that level of technology yet, but people who are familiar with
Excel can now do pretty incredible things with data with a bit of a combination of willpower and
determination, and perhaps most importantly, with the business understanding about what they're
trying to solve for, that will take you a long way in terms of being able to draw out necessary
insights that you're looking for, and to the headline question, maximize the value for the
data. I would contend to you, Eric, that the technology has probably moved ahead of the ways
of working in the people side of things, which means that people are still catching up with what
is possible, and as they catch up, they'll be able to find amazing stuff in the data if they know what
they're looking for. I wish we'd had the technology like we have now when I was starting out in my
career, but does that mean, again, thinking about some of the people you've worked with in the past,
does that mean there's now a bit of a threat, a risk posed to the work, the career of some people
who perhaps did have their foundation because their particular skill was in the technology
of using data rather than in drawing business commercial conclusions from the data?
Well, I think you're still going to require that level of technical skill, right? So, you know, you can go
a certain way with just on business knowledge, but at some time, you're going to have to
integrate the data and think about what it means. So, there's still certainly roles and
activities that will be required by all of those kind of technology people. What they will need to
do, though, I think all of us need to do in our careers is how do we main currency
with the latest technology? How are we using it? Because it is rapidly advancing so quickly,
whilst there may not be jobs that get reduced or made redundant in terms of the nature of them,
the tasks that they do may well change, and if you're not able to, if those tasks can be automated,
and a lot of organizations, you know, look at automation and they look at AI as opportunity to
automate tasks, what are the tasks you're going to be doing in your role? And so, you need to be
asking yourself those kind of questions around what does my job look like? How do I use this
technology to be better in what I'm doing? And how do I continue to evolve the skills that I have
I've got today and build on those to make them a little bit more future-proof?
Mm, adapting and evolving, and that was a big theme in the book that we both
wrote together, if you may remember, in terms of taking the skills that were being used,
but not getting stuck or hung up on a particular objective. Now, if I think about RAFM professionals,
again, sometimes they complain that they lack influence, that they don't have as much
input into decisions being made by the companies they would like. What arguments or methods do you
use to encourage businesses to make data central to the way they make the big and important
decisions? Yeah, it's a good question, a great one. Look, I do think that data has a very high
profile now, you know, at the executive layer, you know, all layers of management and at the board
layer too. So, the fact that a lot of people are talking about it, and you used your quote
kind of earlier as well, a lot of organisations have recognised that to remain a business, not
necessarily even be winning business, they have to start using, you know, the data that they've got
in a more efficient way and an intelligent way, because the days of kind of hunch-based guesses,
or, you know, well, we did this three years ago, so it might work now. Those, I think,
that approach to decision making just feels antiquated in large part. Now, I just want to,
just to be clear, I don't want to give a view that it's all about the data, there's still a huge
amount of space kind of for human creativity here as well. In this process, not all decisions
can be supported by data, but knowing which ones make sense and which ones perhaps may not,
where data adds value is important, the quality of the data on which you're basing those decisions
becomes vital as well, especially if you're making kind of big decisions, whatever big decisions may
be for you as well. But if you're not drawing upon data to support the decision making that you're
doing, then certainly you're probably missing a trick there. And this is, Eric, if you go back,
we want to go back further before even the invention of RA, so it's always called decision
support, and it still, I think, very much remains decision support rather than necessarily leading
only with data. Of course, there'll be some things that people want to automate,
but they should do that with a lot of care, I would suggest.
You make some really strong points, Dan, and this is something that I think doesn't get
unpacked sufficiently. When we tend to talk about data, data science, even data-centric
organizations, there is a little bit of a gap in our understanding, a little bit of a blind spot
in how we think about data sometimes, because when you start talking about data, you make it sound as
though things are certain, that you know how things are, that you can just determine what
has happened, that you can make reliable predictions about the future. But that's never going to be the
case. There's always going to be risk, and that risk needs to be factored into any decision you
can make, because not everything can be known. And even if you've got perfect knowledge about the past,
there's no guarantee that the future will follow what the past has happened. So there's always
going to be uncertainty. How does a data-centric organization, how do professionals who are getting
the most you can get from data, still bring uncertainty and risk back into decision making,
so that you're using data to the full, but you're still being objective when evaluating and allowing
for what you don't know when you make a decision? Well, you're asking how can you be
objective in the subjective areas? Well, I think the first thing is acknowledging the
point that you made, that we don't know everything. The data that we have available to us
may not be complete, or it may not be of the necessary quality. This is where some of the data
management disciplines become really important to inform the decision, or form the basis of
the interpretation of what the data is trying to tell to you as well. You know, most predictive
models will give you an error range anyway, but again, that is based on statistical methods,
and it's very based on the historical view, not necessarily the forward-looking view.
You know, we know the world is a lot more interconnected, and those methods largely
are strong where perhaps cause and effect can follow quite closely, but where those things are
a little more hidden, or a little bit multi-chained, those can be difficult to identify.
So, you know, I think being aware of it's important, knowing that you can actually probably predict
quite accurately, potentially in many cases, but that doesn't mean you know what's caused the
behavior that you're looking at, or the prediction, as well as an important thing to acknowledge. So,
you don't necessarily understand the system any better, you just make an accurate prediction
about where it might go, and sometimes understanding why things happen are probably
just as important about what might happen in the future. Yeah, the deeper understanding that comes
from knowing why things are the way they are, as opposed to just seeing the patterns in the data.
Now, roles like revenue assurance, fraud management, they tend to be focused on operational
risk, like a lot of other data-oriented roles are that involve risk, but the thing with operational
risks is they lend themselves to data in a way, because you're measuring how many times a customer
complains, or how many appointments are missed, how many products come off your production line
that were faulty, and so on and so forth, so they lend themselves to a lot of data analysis.
But in the end, you're only managing then the operational inefficiencies of your company,
you're not perhaps looking at the really big decisions where there could be much greater
potential value, the CapEx, the big CapEx investment decisions. Could more be done,
should more be done to exploit the data to influence those big CapEx decisions too?
Well, if you're talking about CapEx, and maybe I'll think about big investments, maybe in network
or infrastructure as kind of large capital decisions that telcos particularly tend to make,
the answer is yes, but how you might approach that from a data point of view to resolve
those kind of capital questions might be a little bit different to some of the more traditional
methods. I know particularly in the energy sector, and with the drive to, excuse me, to
renewable energy, there's a lot of capital decisions around, you know, what should we do
with our power plants? Should we upgrade them or should we wait for a new technology? Just pause.
Hopefully you can edit that out. And so they are using different kind of analytical approaches to
try and predict future consumer behaviour, not based necessarily always on historical behaviour,
but what future behaviour might look like in terms of uptake of some new technologies,
and you can think about rooftop solar and battery technologies to determine future load
on the power grid. And from that, then they can make capital decisions to say, you know,
for example, in this particular area or geography or postcode, whatever granularity, we're expecting
a high uptake of those kind of technologies. So we're not going to have to generate as much energy
into those areas because they'll be more self-sufficient. And what does that mean for
our capital profile for investing in that particular area? Now, I think the same kind
of logical thinking may well be able to be applied in the telco sense. I know there was some time ago
now, so this is kind of an older use case when 4G and 5G networks were starting to be rolled out.
There were some decisions being made about what kind of, what's the extent of capacity we should
build into the networks, into each geographical area. And there was debates around what will the
future pricing potentially look like to support this? What's going to be the uptake? And so if
you can kind of have models, and there are agent-based models that support these kind of
approaches that can make broad predictions about future consumer behaviour, which we often get
wrong, by the way, in our exponential era. But if you can make those predictions, then you can
maybe more close to what might happen in reality, size your network. And that's your
capital decisions, right? Rather than necessarily putting overcapacity in, or perhaps even worse,
having undercapacity in your network and not serving your customers the way you'd want to.
You make it sound simple, but then at the same time, it sounds so tricky to know how to
begin to unpack this. Is there a particular methodology that you follow in practice?
How much are you relying upon instinct in terms of the subjective part of deciding how to be
objective with data for want of a better way of putting it? Yeah, let me talk generally,
and then I'll see if I can make a specific kind of example in terms of some of the approaches that
we've done historically in the past. So we would see an agent-based model based on the population
census for the country. So if you can imagine the attributes of the population that's gathered
in the census, you create a virtual agent, or just imagine a virtual human being that reflects
the distribution and demographic attributes of people in different, let's work in a postcode
area. So if there's 5,000 people in a particular postcode area, we'll have 5,000 agents that have
the same representation as the census provides us for that particular area. So we'll have the
same age profile, the same gender profile, the same income profile, and the same home ownership
profile, if I'm talking about the energy, for example, everything that census can collect.
And so whilst it may not be exactly, there may not be a person exactly like that, the distribution
is broadly the same. So we've got these agents. Now we've got to decide or determine how those
agents, or you can call them people, but they're obviously all virtual, might respond to different
change. This is where the world of primary research becomes quite useful, because you can go out and
talk to a sample of people, particularly in cases where there isn't the historical
information to base it on. So if I use that example I used previously, we don't have, or
five, 10 years ago, we didn't have that historical base of how many people, what kind of people have
rooftop solar on their houses. So you've got to go and ask people. Primary research becomes
important, but we also know there's sometimes some challenges with primary research and what
people don't always know or want to articulate, kind of what might actually happen in real life.
This is where we can draw on academic research somewhat as well. So you can kind of scour
the academic landscape for pieces of research around things that might be relevant to how
humans might respond in new or novel situations. So you can look, if I used again the rooftop
solar example, what has been the uptake? Who is more likely to take up new technologies that are
likely to have a sustainable benefit? Or who's likely to take up new technology? And then you
can say, well, this may be the wrong conclusion because I can't read what the research says,
but you may conclude that younger people are more willing to take up new technology
than older people. I appreciate that's a generalisation, so we'll just, hopefully we
won't get too much comments back on that. Now you go and feed your agents with that kind of
information to say there's a greater propensity for the younger people in your 5,000 person postcode
to take this up than the older people. And you might feed it more information as well. Home
ownership becomes important because if I'm renting a property, I can't put a solar panel on my roof
because it's not my property. So I have a constraint on that particular agent. So even though I may be
a younger person, if I'm renting, I still may not yet put anything on my roof. And you can progressively
continue to build this up and build this up to some understanding. Now you can start to factor in,
because you're going to run this model over multiple time iterations as well. So you can
build up a social factor in there as well. The kind of things that we know happen in the real world.
Like if three of my neighbours have all got this thing, that may influence my decision as well to
want to get kind of rooftop solar. And so that can be influenced into the model over time. And if I
run this model over multiple iterations, i.e. kind of timestamps, I can start to potentially
get kind of closer to what the future behaviour of these agents might be. And then I can kind of
make an informed decision about how I might respond in these situations. And we have seen this.
It has been used, as I said, in the energy sector in particular. I'm not aware of it being used
in telco. To date, it has been used as well in government and social services around how you
might nudge populations to do things as well, and where you might get the best effect of nudging.
And that's a whole different area around behavioural economics. But that kind of application
of data analytics is probably something new and emerging. I guess we're seeing, particularly,
as I said, where there isn't that historical basis of data to draw your conclusions from,
where AI in those areas has been so strong, different kind of problems, and where there's
more complexity in the human response, and where human response may not necessarily follow a
logical path, lead themselves to that kind of approach. That was a bit lengthy.
It was music to my ears. It was bliss to listen to you. I have to say,
I could talk all day with you, Mike. It's so intriguing, so fascinating,
what you're seeing in the data, what you're gaining in terms of your work. And I can feel
your passion very much, that you're enjoying your work as much as you ever have. So thank you for
sharing that. We've only got time for one last question. And this one is an impossible question,
but I have to ask it, because you brought up AI a couple of times during the course of our
conversation. Everybody, of course, now is talking about AI, to the point of insanity. You
can't escape people talking about AI, since chat GPT made everybody realize that we are entering
a phase where nobody can perfectly predict how work will be altered by leveraging this artificial
intelligence. Is it possible to make any worthwhile predictions about the way AI will change
how businesses will operate? You are in the business of trying to anticipate the future
at EY. What can be said? What can you give us? What tips can you give us for how you expect the
world of work to change as a result of AI in the near future? Well, at least you said near future.
Mind you, if you said distant future, then I could say something and no one would come back
to this point, this instantaneous time to tell me I was wrong. No, no, I'll be watching this again
in 10 years time to check up on you, Mike, to make sure you've got your predictions right. Don't
worry. We'll get you back on the show in 10 years and review this performance. See if it's right or
wrong. Look, I think the application, this is my kind of personal view only, right? I think
generally the application of AI to date has been narrow and for specific kind of problem solving.
So it's very good at solving kind of very specific areas that you might look at when
you've got it targeted at the right issue. This is why we don't have generalized AI and
terminators kind of running around because they can't solve multiple problems as easily.
The rise of chat GPT and kind of that generative approach, that's interesting. I think I did,
and I would love to attribute this, and maybe I'll come back to you to attribute this, but
somebody said it to me as like a JPEG of the world. So it's a little bit fuzzy on the outside,
but gives you a general view of what things look like. So it's not, it doesn't have the precision
necessarily of a human decision maker yet, albeit it's learning as we all know at a kind of crazy
rapid rate. Look, I expect in the near-ish future, more of us, more people will need to
work with kind of AI algorithms to make better kind of decisions or more informed decisions.
That'll be a reality for increasing amount of people. However, much of it may in fact be
transparent or invisible, sorry, rather to them in terms of how they're operating.
Yeah, I mean, when you and I go on a maps application on our phone to try and find our
way from A to B, we're not exposed to all of the intelligence underneath the hood that helps us.
We kind of just get the answer and kind of off we go. And I think that's probably going to be
the experience for many people in terms of how they might engage without even knowing what they
are, as they do today in many of their daily interactions. It will change, no doubt, it will
change the nature of the work, it will change the kind of decisions that people might make versus
what organizations might be comfortable are made in a more automated kind of way. But there'll
still be roles, there'll still be role for humans there, particularly where there is a human being
affected by this. And this is why the ethics of AI is receiving such attention as well around,
you know, should we even use it for this purpose? Is that in line with our organization's
ethical frameworks, many of which may not yet have been established and are kind of being worked on
to know what they believe is permissible and acceptable kind of within their context as well.
So look, the short answer is yes, yes, it will change. I think, you know, you might find people,
they don't manage people, some people won't manage people anymore, they'll be managing,
you know, an army of AI agents operating across the business and making sure they're still
performing to their expectations. But it's promising, but it's also scary. And I think
most people acknowledge both sides of that coin that it needs to be looked at. And, you know,
come back to your starting point around where risk professionals can be involved. And as we said
earlier on, you know, how is your organization using kind of these tools? And are they using it
in a right way that actually supports managing risk rather than perhaps creating more risks
themselves? Absolutely. Time for the risk people to step up and be counted because they cannot
allow this topic to just run ahead because of the seriousness of the issues. Look, if it makes me
unemployed and I'm out of work, I'll come down to New Zealand and we can have more conversations
like this. We're out of time, Mike. But thank you so much for your insights today. I really
appreciated it. And I appreciate also you probably your kids are grumbling right now wondering why
their dad hasn't made them breakfast yet. So thank you again for your time, Mike. I'll look forward
to our next conversation. Thanks, Aaron. Always a pleasure.
Many thanks to Mike Willett there for that fabulous interview. Guys, I wanted to get your
views there a little bit on Mike's comments on the end about artificial intelligence.
Starting with you there, Lee, do we think that generative AI may have even more impact on
discovering new ways to increase profits? Or is it necessary that we should make a bit more effort
to talk of human intelligence to make sure we don't all lose our jobs? Well, we're already seeing
AI, which is being used in telcos. It's used at the moment for monitoring the network. We use it
in chatbots, in customer services. We actually spoke earlier, we actually use it in detecting
patterns in fraudulent calls, especially Wangarian spoof numbers. We also use it for blocking
suspicious SMS messages as well. And also start to see the marketing teams using AI to detect
customer sentiment. But I think the list is going to grow, but I think we're still some way off the
singularity. Singularity, of course. Ed, I've got to bring you in here. I don't want to see a
singularity, Ed. I don't know how you feel about the possibility of a singularity here. Are we
missing something in terms of how we measure and understand artificial intelligence versus human
intelligence? Are we actually appropriately weighing the two up? Are we tending to perhaps
exaggerate, put too much emphasis on artificial intelligence because it's something that's being
sold, whereas human beings, well, they have to sell themselves and with the same old product we
always were? Yeah. I mean, it's a confusing subject because there's just so much hype and
noise around it right now. I like what Lee had to say because those are the kinds of
practical use cases where it seems very applicable that people that know what they're doing
and have an engineering-led approach to it seem to be applying the technology effectively. And
that's consistent with people I speak to in the NTM forum and in service providers as I'm
doing the research that I do in general. So all that kind of stuff makes sense to me. I think it's
sort of my caution as to the greater market because of the amount of hype around AI right now.
And I do feel a little bit like there's this effect as we were chatting about that it's like
the conquistador helmet found by the natives. And half of them are screaming that it's a god
and the other half are screaming that it's the devil and they're fighting over it. And that's
pointless. I just feel like that's the public discourse and the industry hype discourse around
it where really the more effective discussion is around the kinds of things Lee was talking about,
about ways that we can improve security, improve how data is used, improve how networks perform,
especially when you're talking about critical services. All those kinds of things make sense.
There are a lot of cases when you start getting into really complex orchestration
where it starts to make sense that there are things that AI can help as a tool in order to
help solve some problems. But I think that when we then get into that next level where people
maybe don't understand it as well, start wanting to make rules, make laws, make suggestions of
application for AI to things that may not be necessary, it may not be the right fit. But like
we said, you feed into that hype cycle again. I think that's actually where the AI danger starts
to come from. Even if it's just because it's the next big tech con, then whatever the big AI brand
is turns out to be the next big fake tech company. Even if it's just that and a bunch of investors
get burned, that's bad enough. But I feel like it's on track for something like that to happen.
And that's unfortunate because again, it kind of covers up the good practical uses of this tool
for pragmatic purposes, other than just trying to sell me more stuff.
I hear you. I hear you. Well, I'll just read out one last question from the audience here
before we call it a day on today's show. Anonymous viewer here says,
there's a thread in science fiction where thinking machines get banned or destroyed.
Does the panel ever worry that this will become true? Let's start with you, Ed.
It's funny because that was like one of the first conversations I think, like lengthy
conversations you and I ever had. It was about the Butlerian Jihad in Dune and that exact concept
that you could eventually get to that. I don't know. I'm actually more concerned about the other
way around that people have become so dependent on the smartphone or whatever machine they're
interacting with to help them do their thinking. That's where I think that we become more of a
danger to ourselves because of the technology. And then it becomes, you're not going to destroy
the technology because you couldn't possibly live without it. You wouldn't know how to
find your way to the restroom.
Lee, can I interrupt you, ask you to get off your smartphone there for a second and
I'll answer this question about whether people are using their smartphone too much.
It's happening right now.
The reason why I was on my smartphone, I was trying to Google how 9000 look. I think we don't have to
worry about how just yet. I think we're a long way away from that.
We'll leave that as the parting comment for today's show. Thank you both of you guys.
Fantastic show today. Thank you for everyone who watched. I'm sorry I didn't have time to
read out as many comments as I would like today. That's it for the penultimate episode in this
season of the Communications Risk Show. Join us next week for the season finale when Ed, Lee and
I will discuss the highlights of this season and map the major risk topics that will dominate as
we enter 2024. We'd love to have your input in helping us understand what you think are the big
risk topics that we need to be discussing more often on this show. The live stream for next
week's show begins on Wednesday, November 1st at 12 noon, US East, 4pm in the UK, 9.30pm in India.
Note the change of time from our previous slot for some regions. That's because daylight savings
is about to end in the UK. As I say every week, you can spare yourself the trouble of calculating
offsets between time zones by visiting tv.commsrisk.com and clicking the link to automatically save the show
in your diary at the right time zone for you. And whilst next week's show will be the season finale,
it is a good idea to subscribe to our broadcast schedule so you will have next season's show
added to your diary automatically. Thanks again to today's guest,
Mike Willett. Thanks to my co-presenters, Lee Scargall and Ed Finegold. And thanks to the
hard-working producer of this show, James Greenley. I'm Eric Priezkalns and you've
been watching episode 10 of the second season of the Communications Risk Show.
All episodes can be replayed at this show's dedicated website, tv.commsrisk.commsrisk.com. Be sure to
catch up with news and views about risks faced by coms providers and their customers at our main site,
commsrisk.com. And don't forget the free resources for risk and assurance managers
at riskandassurancegroup.org. Thanks for watching today's show, we'll see you next Wednesday.