Artificial intelligence will revolutionise global business, but how should industry stakeholders approach this complex and often misunderstood technology?
SBC speaks to Qubit CEO & Co-Founder Graham Cooke on the current context of AI and how to gain inclusive learnings on the subject matter without distractions or disruptions.
Leading Qubit, Cooke has worked and consulted on a number of ‘Tier1’ industry Big Data and digital personalisation projects for clients including; Ladbrokes Coral, Betfair International, Sky Betting & Gaming and BetBright.
SBC: Hi Graham great to catch up, as a technology and data veteran, what draws you and Qubit to AI advancements and innovation?
Graham Cooke: Firstly, I think it’s important to make a quick point that most of the concepts in the AI field, such as machine learning and deep learning, have been around for years. However, we previously lacked the economically and technically viable environment in which to apply these concepts at scale.
Today the combination of increasingly powerful computers, decreasing data storage costs and the cloud computing delivery model has sparked a ‘renaissance’ period for AI.
Pursuing AI was always a natural fit for Qubit. We’re a personalisation-focused business and the only way to deliver personalisation at scale is via AI. Furthermore, we knew we had the ingredients to push the boundaries of the technology. AI is only as good as the data you feed it, and our data pipeline processes and stores billions of digital interactions on behalf of our customers. You can only describe a system as a pipeline though if the data has a destination (or purpose). Qubit’s destination is digital personalisation and we always knew that whilst rules based personalisations are a great place to start, AI-driven personalisation would take us to the next level of scale.
SBC: At present, AI has taken on a broad definition. As a technology leader, how do you look to specifically define effective AI innovations and processes?
GC: AI is taking an increasingly broad definition, partly because new applications for AI are constantly being developed. However, there’s also an alluring, almost mysterious implication in saying that your solution or product is ‘powered by AI’. We’re seeing many features haphazardly painted with the AI brush for marketing reasons, even if they’re actually just traditional algorithms.
At Qubit, we focus mainly on machine learning, employing algorithms that can learn from and make predictions on data in order to automatically improve programmes. For example, we use this in eGaming to spot key signals in customer behaviours, uncovering where the biggest opportunities for improvement may lie. The machine learning engine can identify these opportunities instantly, but it would take humans months to sort through the same data.
Ultimately any application of AI can only be classed as effective if it improves your business KPIs and if you can prove that this uplift is caused by the AI element. Despite all the recent developments in AI, this is still my definition for effective use of the technology.
That being said, I have noticed that we’ve become more patient when considering whether AI is effective. A Machine Learning algorithm needs time and data in order to learn and improve. The more data you feed the model the quicker it learns and so many organisations are learning that patience is required. It’s no use writing off a product as ineffective when it hasn’t been given time to improve, and the history of AI is littered with success stories that started out as abject failures.
SBC: In the industry’s current context we are seeing a huge emphasis on producing omni-channel dynamics, bridging gaps between land-based and digital capabilities. How do you see AI changing these dynamics?
GC: So much of the experience in land-based operations, whether in casinos or bookmakers, is dependent upon human interaction. In the early days of eGaming, this human element was completely lost. Whilst this may have been a good thing for a poker player with an obvious physical ‘tell’, for many people it made the experience less sociable and enjoyable. We’re now seeing companies striving to curate ‘human’ elements in their digital service, and AI-driven personalisation, which provides tailored and relevant experiences for each user, forms a big part of this.
The emphasis on omnichannel strategies makes commercial sense. Ladbrokes has been at the forefront of this, promoting their multi-channel product during Euro 2016 and welcoming more than 13,000 new active customers over the course of the competition. Ladbrokes has found that retail customers who converted to multi-channel continue to deliver more value than digital-only customers.
We’re also seeing efforts to try and treat customers more consistently across digital and social. However, while digital identification is easy, most land-based business is anonymous. I can see why the operators would like to change this, particularly U.S land-based business closely following developments in digital regulations.
Looking into the future, the latest iPhone developments around facial recognition could potentially help here. If a user is unlocking their eGaming apps via facial recognition, it’s feasibly possible to match this land-based business using CCTV auto-recognition. However, to me this still feels like a bridge too far, and I’m not convinced that either regulators or players are quite ready for that level of ‘know your customer’.
SBC: You have detailed changes at a consumer level, however from a leadership perspective, how will AI factors influence governance and future decision making for all industry stakeholders?
GC: From a leadership strategy perspective, AI will naturally accelerate the decision-making process. A key strength of machine learning is the automation of the market segmentation discovery process, identifying priority customer groups for targeting, highlighting the monetary opportunities they represent, and creating new clusters of interesting potential customers for attention. This is all done instantly and without error, and so business strategies can be adapted very quickly indeed.
With regards to governance, there is clearly a lot of talk around data protection regulation, which is clearly very important. However, there are also opportunities for AI to play a role in safeguarding potentially vulnerable customers. Behaviours commonly associated with problem gamblers can be analysed and flagged to operators in real time. AI will benefit the eGaming industry enormously and I believe it will provide even better, even more responsive services for eGaming customers too.
SBC: Moving forward how would you advise to industry stakeholders to approach the discussion of AI with internal teams, in what can be a disruptive and often misunderstood subject matter?
GC: Businesses need to understand what AI is and why they need it. As you say, it’s a misunderstood subject and it’s easy to imagine a generic AI ‘brain’ that can be applied to any problem, magically self-optimising websites and engagement wherever it’s deployed. In reality, algorithms need to be specifically calibrated to function effectively. I don’t think industry stakeholders need to be experts in AI, but they must know where their business can benefit from it, and internal operation teams will be best-placed to point out challenging areas such as missed opportunities to drive additional revenue.
AI should be broached internally as an opportunity rather than a risk. Machine learning eliminates the need for labour-intensive, manual data collection and integration. This means that internal teams previously tasked with this will have far more time to dedicate to those areas that humans excel in, namely creativity and strategy.
SBC: Finally, In your view, has the industry been enthusiastic/ambitious enough when experimenting with AI? How does it compare to other sectors in this regard?
GC: eGaming has been particularly enthusiastic in experimenting with AI, delivering more personalised service at massive scale breeds loyalty in a market with very little. That being said, eGaming faces specific challenges. Harnessing AI to predict what might interest a customer is more difficult in this industry than any other. Big operators will have over a million potential options available to customers on any given Saturday, most of which will be gone by Sunday.
Additionally, AI is often used in Retail & Travel to make tailored recommendations and help people discover new products. In eGaming this process is fundamentally different and product discovery is nowhere near as effective since customers already have an intuitive knowledge of the selections that will be available. Instead, our eGaming AI techniques focus on delivering improving navigation by predicting which markets and sports a user wants to get to, as well as deciding the next best action once users have placed a bet, surfacing other options that may tempt them, such as previous winners.
Both of these areas have helped eGaming really push the boundaries of AI on mobile in particular. Whereas the mobile experience in retail or travel is still somewhat behind the desktop, eGaming is often able to provide a seamless, tailored mobile experience.
We’ve found that there’s a bigger emphasis in eGaming on the operator owning the IP and algorithms to ensure differentiation from competitors. When Qubit develops an effective predictive model, we work to integrate that model into the client’s product and provide the data to train and constantly improve that model. This means that gaming/betting businesses often have a better grasp of the capabilities of the technology and can naturally be more ambitious in pushing the boundaries of AI.
Graham Cooke -Co-Founder & CEO – Qubit