Artificial intelligence and big data consulting: doing black magic with the world’s ‘new oil’?
Artificial intelligence and big data consulting: doing black magic with the world’s ‘new oil’?
Lead Data Scientist at BCG Gamma consulting and Cambridge MBA alumnus, Iman Karimi (MBA 2010), talks about big data, advanced analytics and artificial intelligence and its impact on society and companies. A conversation with Cambridge MBA Executive Director Conrad Chua in the latest episode of podcast ‘Changing Careers’.
Data is the ‘new oil’
We are currently living through something like a gold rush in the sector of artificial intelligence and big data. Every company is using it to harness its power and potential, even to the extent that people are talking about an economic or technological war between the US and China being fought on this front.
Conrad Chua explains in his regular podcast ‘Changing Careers’ that, “unlike the previous IT revolution we largely don’t see the impact of artificial intelligence on big data yet, they are ubiquitous. From the way the latest iPhone recognises your face to Amazon’s Alexa answering your questions. Yet questions remain about how companies use data and AI and what will the impact on society to help us better understand these issues.”
Based in BCG Gamma’s London office, Iman Karimi concentrates on problems related to predictive analytics across a range of industries, often mass consumer in focus, such as insurance, banking, telecommunications, retail and so on.
Many consulting firms can be seen as traditionally monolithic and homogenous in nature, but BCG Gamma is an operational practice-based solutions company. They have quite a bit of autonomy over the recruiting practices compared to a traditional global consulting firm. “So, candidates still do the normal case interviews, but we now apply data science and artificial intelligence as an additional tool,” Iman explains, “We solve problems using different type of tools but in a way that we can clearly explain to the client. This is the most important difference between the job of the consultant and say someone else. We are not successful if we cannot explain how we have solved the problem and ensure that the client buys into that. At BCG Gamma we are not trying to be there forever applying advanced analytics, our main value proposition is that we will enable you.”
We are not going to just sell you fish, we are going to teach you fishing
Across BCG Gamma they use technical interviews, mathematical
skills test and advanced analytics knowledge. There is not an expectation that
each applicant would know everything, but there needs to be some identification
of strengths in some part of the spectrum of some of the capabilities that are
needed at the company. Some other tests are purely hands-on such as modelling
Consultancy skills are still required across the company, in
order for those on the teams to explain the outcome and the complex modelling
and coding to the client at every stage. “How can I show you, to gain
confidence that this is not just black magic but that it is real and it will
help your business if applied?”
“We divide a problem into three main sections,
envision, AI activator and enabling.
“Envision is about thinking through the solutions and
enabling is about scaling. Our job is not to develop an algorithm to address
the problem. Our job is to solve a problem and to solve a problem with AI means
it needs to be incorporated and integrated within your organisation. The AI activation
covers various phases, to identify that applying AI is going to bring value,
then if this is identified you develop a prototype phase, after this there is
incubation to deploy the algorithm in the field; how the model performs in the
real world, there is some retuning and refinement of the algorithm. The final
stage is the integration across the organisation.”
What are the main challenges that clients face when adopting AI solutions
to their business?
“One challenge is simply time – that a climate for
specific technological advances can have changed dramatically during the life
of a project, so for example over the average two-year period of a typically
large IT project, by the time it is complete it can be already outdated.
“The other challenge is that we have hired this group
of ‘geniuses’, but we are not seeing any impact and sometimes the things that
are developed are very cutting edge but either they are not really leveraged
fully or the business has not really seen the value to engage with the solution
or that it is not applied on platforms deeply enough across the company for it
then to have meaningful impact.
“Buying into the ‘black magic’ – there is now a big
market, some are ‘snake-oil’ merchants and some of them may be real and they
say, ‘give us your data and we will give you the results and we have this
fantastic solution, and it has worked for many industries and it can solve all
your problems’. While it is tempting to buy into this quick-fix solution, the
danger is that it does not factor in all the idiosyncrasies for any particular
industry. It is very difficult to apply one algorithm that is going to be a
panacea for all problems.”
Iman Karimi started out doing a PHD when big data and AI
were in their infancy and when data was really incredibly sparse in relation to
a lot of the world’s problems. But having worked in an era of ‘sparse’ data his
background enables him now to view big data through this lens. “Even in
today’s climate there are always some customers and some areas of any business
that you have very little datapoints for”.
“I always believe that mathematics is at the core of
everything. One piece of advice I always give is, ‘knowing how to code and use
packages and so on is very good. But never forget that at the heart of these things
is mathematics.’ If you don’t know what the package is doing you will turn into
just an operator. Then you are likely to make mistakes and you may not be able
to spot it as you don’t know really how it should be.”
Will AI replace humans in many jobs. Will AI replace a large section of
“Like any other technological advancements, it will
certainly change function. There will be some jobs that will be obsolete and
there will be new ones that would be created. It will certainly replace some
people; across Industry 4.0 and some manufacturing sectors that has already
happened. With the consultants the skills that would be required to do the
strategy jobs in say ten years would be quite different from now. Those that
can adapt themselves can still be relevant and in the game. Skills that have
100 per cent overlap with developments by the machines would most likely be
“Since the industrial revolution two centuries ago,
humans have constantly been afraid that the machines are going to replace them
completely. It might well happen someday.
“Creating something as sophisticated as humans that can
adapt itself to different situations, think about different problems at the
same time and with different type of environments, that is quite difficult.
“Game of Thrones fans will be interested to read that
there is a version of the final book that has been written entirely by an
algorithm. The algorithm has digested the whole series of books and then has
tried to write what would be the sequel of that. But is it going to be as good
as something that the author writes?”
There is a lot of public anxiety about AI not just about the
impact on jobs but also about transparency and fairness. So, for example with
regard to credit approval, which is now done almost entirely using AI. Would
people accept the decision by AI more than they did in the past from a human?
But there are still a lot of questions unanswered. Even for
companies, AI and advanced analytics are still to a large extent like ‘black
“They don’t know exactly what it is and as it becomes
ever more important, so it cannot be completely ignored anymore. What are they
capable of and what are they not? Education on the capacity and capability of
this sector will help reduce that anxiety.”
When the internet first came along it reduced the symmetry between companies and so for example prices could be compared across suppliers. Now, for a while at least, AI can give a company the advantage again, they are able to put the resources in to digest the data and place themselves in a more advantageous situation. But it cannot be in an abusive way for a long time. With GDPR regulations and the inevitably further regulation, specifically to the derivative information captured, for example. “Data is like the oil of the future, but the main criteria is about trust, and that is why scandals that companies are currently undergoing with breach of data, will be very damaging”. If people stop sharing their data, and legislation and regulations stop companies obtaining data, your algorithm is only as good as the data available.
“Personalisation of data is of special interest to me,
what is the right action at the right time for each customer. If it is done
right it is going to benefit both sides, not a ‘win-lose’ situation towards the
company but a ‘win-win’ situation. I get what I want, and the company gets more
value out of me. If I have to sacrifice a little of my privacy the value to me
is worth it as I am now receiving the relevant recommendation and the service
that I needed and so on.
“Before taking my Cambridge MBA , I had stumbled on an article on artificial intelligence and I decided to focus my masters’ thesis on neural networks and AI. I became fascinated with the use of artificial intelligence on management and decision making. The most important decision for me was to stick to my guns and so my PhD topic was then on uncertainty in decisions and risk management specifically when you have sparse data. I then moved through analytics and the insurance sector and I only came to the role of consultancy in 2012 when big data and the role of analytics had finally become popular and big data broke through.
“If you stick to your beliefs as long as they survive
the test of reality. This revolution was going to happen sooner or later and by
reinforcing my knowledge in this field I knew it would be valuable when it
finally came to pass, and it became true.
“During my Cambridge MBA it was interesting and fortuitous how faithful I remained to my statement of purpose.
“If, by the time I finished my MBA year, the climate
wasn’t there to take AI and advanced analytics seriously then I was going to
start my own company, but by the time I had finished it was already starting.”
Conrad Chua concludes with a recommendation to, “think about which part of your job is routine and vulnerable to replacement by AI and machines. While it is uncomfortable to think about this, as Iman has shown, you can always make a career transition to stay ahead of those trends.”