Welcome to the Executive Data series.
DXC created this new program to provide advanced insight into the data domain. In a series of conversations, DXC experts will explore data-driven decision making, and offer their perspective about what it takes to be successful in data, information and knowledge activities.
Mohammed 'Khal' Khalid, global advisory director at DXC Technology, will moderate our series. Discussions will draw on research conducted by DXC and upon our executives' experiences working with customers.
In this first conversation, Khal welcomes Director of DXC's Global AI Practice Kal Kanev. We invite you to listen to their full conversation or, if you don’t have time now, to a short extract about how an “AI moonshot” can propel high-value, high-impact AI projects. You can also find a full transcript of the discussion below.
Kal Kanev
Read Kal's new perspectives:
The conversation
Q. Welcome Kal. Why don't you start with a quick introduction?
A. I am the director of the Global AI Practice at DXC Technology. I spend most of my time helping our global customers with their AI journeys, being a strategic advisor for the C-suite and also directing AI teams in successfully embedding AI. I've been working in the AI/ML domain for over 16-17 years now, and it all started initially in the medical robotics lab at Johns Hopkins University some time ago, where I was helping design computer-assisted surgical instruments, and that unlocked my passion for AI. So fast forward to today, I’m mostly focusing on helping DXC customers scale AI and ultimately change their business.
Q. Let's talk about the business need and challenge. What are business leaders saying to you around AI? Why do they want this stuff?
A. Businesses and business leaders need to change and adapt. So they are investing in AI because they want to grow by reinventing or bringing brand new products and services to the market. Some of our biggest clients are actually revolutionizing their whole portfolio because they want to be more of a digital company and offer their customers new, valued experiences.
Another reason for investing in AI that I hear from business leaders is because the ecosystem that businesses operate in today is changing. They're generally asking about guidance on what to pursue with the help of AI and how. Some clients want to be on the cutting edge and go for the high-impact AI moonshots. Others take more of a conservative approach and want to implement AI where it is a bit more mature, where there is some successful implementations already being made in their industry. So they want to follow more of a well-established path.
Q. When you're helping people to identify and act on the high-value business cases, what does success really look like for them?
A. For some, success is optimizing and automating to save costs and do things better, faster, cheaper. For others, it is being ahead of the competition and having unique new products and services. What I tell customers is to think about AI not just for automating and optimizing, but to use AI to drive growth. And success usually, in the most pragmatic sense, means having ROI and winning in the market.
Q. So let me ask, what do you see as the key to success? What approaches do you think work best when progressing one of these focused high-value business opportunities?
A. There are multiple aspects to this. The key is to have a clear strategy for AI, specifically how it ties to your business and data strategies. So having a path to monetize your data and infuse AI insights into products and services is crucial. It is very important that the business drives these initiatives so they don’t become an academic exercise for the data science team. And speaking of teams, having a proper AI team is definitely the key to success. That includes data scientists, but also data engineers, application developers, business AI leads, AI architects. The AI architect, for example, is a key role that truly understands the business and knows what can be done with AI to help scope and provide focus on the high-value, high-impact initiatives.
Q. Given all of that and, again, based on your experiences, what are the common pitfalls that you observe? Ultimately, the question I really have here is what should CIOs/CxOs avoid when embarking on a new initiative?
A. One common pitfall is not being clear on what problem or question AI will address for the business, specifically a particular business unit or part of the organization. Not having enough high quality data is another pitfall. Businesses typically have a lot of data from different systems, which needs to be onboarded and curated, but whether that data is the right data is usually an afterthought. So getting your data right and having that data foundation to do AI is a prerequisite.
Another oversight is having a team which is incomplete to address an end-to-end business solution. What I mean by that is, usually companies would hire a number of data scientists – some PhDs even – thinking that if they hire very smart people that can do the machine learning, they would be positioning the organization for success. This is part of what’s needed, but it’s not all of it. In reality, that does not quite work. There is actually a need for a team that has various other skills. And the business should also be part of that team, which in turn suggests that there needs to be, to a differing degree, AI skills across the company.
To answer your question specifically, CIOs should strive for AI teams that are comprehensive and comprised not only of data scientists. There is the need also to overcome the “POC [proof of concept] trap,” which is avoiding projects getting stuck in the pilot phase. To that extent, CIOs also need to think of a proper strategy and an approach for industrializing AI efforts to see the benefits. Otherwise, things just remain as a technical experiment and the business is not able to benefit from AI as a technology.
Q. So what I'm hearing is that there needs to be a multi-disciplinary team; it needs to be contextual to the type of decision that's being made. You mentioned co-innovation. I would love to hear more about that. And what else do you think is the key to a successful AI initiative?
A. For less mature companies, the simple advice of starting small and building capabilities over time is usually the best advice. For some more mature enterprises, however, just completing POCs and pilots is not enough. At DXC we see the approach to be one of the most crucial aspects to successfully implement AI. We learned that it works well if we do co-innovation together with clients and follow either a framework that we have for AI moonshots, or take the design thinking and ideation route as the starting point.
Q. I hear a lot about design thinking, so let me start there. Why is design thinking so important?
A. It is important because it ultimately allows us to do a different kind of design and engineering for AI solutions. We want to really understand the business, we want to understand the users and, at the same time, we want to be able to challenge some of the assumptions that are out there. So it gives us the ability to redefine the problem space. And the goal is to identify alternative strategies and solutions that are not necessarily apparent to the AI team and the wider business team. Design thinking really is a great way to involve a wider team of subject matter experts and users to take the opportunity to – in an agile way – address the solution and implementation of such a project.
Q. And what about AI moonshots? Sounds really interesting, tell me more.
A. AI moonshots are innovation projects that are usually a catalyst for change. They distinguish thriving companies from the ones that only focus on sustaining the business. We see that moonshots usually stretch teams to do something a company has never done before, be it a product or service. It is important to have the setup to do such projects in a way which allows for experimentation and, at the same time, accommodates for failing fast and failing cheap.
At DXC we have an AI moonshot framework which we apply as a best practice to help our customers in co-innovation projects. And the reason we want to focus on co-innovation is to come together with our clients to help them solve for their competitive advantage and to generate their own IP, and truly lead in the market. That’s truly enabling them for growth.
Q. So, when folks take the advice that you and your team are able to provide, what should they be thinking about next? Simply put, what’s next after a successful experiment?
A. For companies that are a bit more mature, they need to think as a next step to have an industrial-grade AI platform, and think about having the AI solutions in production from the start. In other words, be ready for “production by design.”
Solving and eliminating the operationalization challenge – that is, getting things in production – is probably the most important bottleneck that needs to be addressed.
More AI-advanced companies usually focus on empowering their power analysts, and what we call citizen data scientists, to do AI, which is great. That's how you scale. So it's not just the PhDs that we've hired that can do AI, but we use the rest of the talent pool. And to do this, those companies need to simplify the ML foundation and the operations. They need to have everyone working in the same platform – from incubation to production – and following a series of best practices. All of this is commonly known as MLOps. DXC has created, for example, MLOps quick starts to enable clients in a matter of days to have such an industrial-grade platform so they can focus on the engineering and customization of AI, not so much on crafting and perfecting the ops associated with AI.
Q. Finally, how should people get a hold of you if they want to discuss this further?
A. If people want to reach out they can connect with me on LinkedIn. That's one option, or if you're already working with any of the DXC account execs, feel free to reach out through them. You can also fill in the contact form on our website.
Listen to more in the Executive Data Series
The value of procurement data (Session 2)
The banking customer in a data-rich world (Session 3)