Leveraging AI in the business
While we tend to think that AI is a new technology, the concept has been around for many years, and its impacts are already well established in the enterprise.
One of the earliest examples of AI in Africa is USSD on mobile phones, where businesses used programmatic language to respond to end-users, giving options and then providing an appropriate response based on input received.
So says Douglas van Wyk, regional manager at Infobip, adding that, contrary to the Hollywood-informed perceptions, in reality, AI is not necessarily a walking, talking robot. It takes many forms and has many use cases and can be simple or complex depending on the application. Currently, AI is commonly used to take mundane, repetitive and low intelligence tasks away from human agents, automating them and vastly improving efficiency. This is especially useful for static information such as location data and frequently asked questions, which can easily be offered to users in a self-service manner. It is also used in predictive analytics, something that is valuable for solutions such as customer engagement.
Paula Sartini, founder and CEO at BrandQuantum, says technology advancements and adoption are driving customer expectations, which are trickling from traditionally non-IT industries into a broad range of sectors such as real estate and financial services. AI is transforming the way in which companies do business, with several departments reaping the benefit of AI solutions. In addition, while AI and automation offer time- and cost-saving benefits for companies, the real benefit of these solutions lies in the ability to improve customer services. Customer experience is driving digital transformation and the customer should be central to every technology decision a company makes.
“However, the technology solutions should not replace the human experience and these technologies should remove repetitive functions from employees so that they’re able to focus more attention on being strategic and meeting customer expectations. Equally important is proving tools and solutions that empower the entire organisation to deliver a consistent brand experience across every department,” adds Sartini.
Exponential volumes of data
According to Glenn Noome, director at Smart Integration, AI and IoT go hand in hand. “Over the next few years, we’ll see billions of devices added to the internet, creating exponential volumes of data, too much data for us to analyse and use to make proper business decisions. AI can assist here, by analysing all of this data and helping make informed decisions based on the best outcome. A simple example of this is AI built into CCTV cameras. Instead of an operator trying to focus on multiple cameras, AI will monitor footage and advise on pre-programmed events. In addition, AI could be used in law cases, where data from previous cases is analysed instantaneously and give the correct outcomes predicted. Furthermore, and importantly, in the field of medicine, robotics could potentially perform certain operations under certain circumstances using AI.”
AI offers significant value to businesses, adds Kevin Dherman, SYSPRO’s chief innovation officer. “It eliminates repetitive tasks, better predicts trends, and can take action before problems occur. It also enables businesses to provide a consistent customer experience. AI and machine learning (ML) give computers the ability to make sense of and learn from data to perform specific tasks without manual interference. AI provides analysis and insights to users, addressing the large amounts of structured and unstructured business and industry data that companies increasingly need to consider as part of their decision-making process. With AI in place, interactions with customers will move from straightforward transactional models to multidimensional conversations spanning a variety of complementary channels.”
When analytics is infused with AI, businesses can start to truly enhance productivity among the human workforce, adds Clinton Scott, MD at TechSoft International, adding that when analytics is augmented with AI, to automate tasks, improve workflows, and discover insights, everyone is empowered.
"Knowledge workers are already comfortable with the use of AI-led technologies and insights as a means to improve day-to-day non-technical business functions. AI is being used in financial services chatbots, in call centres and even in ad-serving software when transacting online," says Scott. "But we are only scratching the surface of how AI can help when augmented with day-to-day analytics; AI can help fuel and lead massive productivity gains in business. The argument for AI-enhanced productivity in the human workforce is not a new concept, nor is augmented analytics. But they are still widely misinterpreted by businesses that are still unsure as to how best to deploy them in their organisation, or who still believe they should only be accessed by the data experts. By its very nature, augmented analytics is defined as the ability to enable technologies, such as ML and AI, being applied to data generation and, in turn, being used for better insight generation and explanation. In short, it augments people's behaviour with data, how they use and explore it, and pull this through to BI and analytics platforms. Now, when it is coupled with AI-infused analytics and ML technologies, that is where the real magic starts to happen. Augmented analytics was cited by Gartner as a top strategic trend for this year, and as harnessing and driving innovation in business. Where it starts to work is when it’s used to augment the intelligence and behaviour of multiple users to create automated insights that can then be copied and carried over to business functions, ultimately guiding favourable business outcomes.”
So where to begin on an AI journey? For business, any AI implementation should always begin with the customer experience in mind, says Van Wyk. When you adopt this approach, the only limitations for the application of AI are human imagination, and often, more importantly, business budgets. The first and arguably the most important step in any AI implementation is to have a solid understanding of business processes, available resources, business requirements and customer needs. Once you have this, you can identify areas that can be adjusted and streamlined using AI. It’s important to remember that AI is not a one-size-fits-all approach – not every business requires 24/7 communication channels, for example, nor do all businesses require a chatbot solution. For some businesses, an SMS offering might be the most appropriate, given the needs and preferences of their customer bases. For others, webchat apps might be apt and well-received. At the end of the day, AI works best if it is channel-agnostic and offers full two-way communications. This is essential to ensure adoption rates are high and return on investment is maximised, by streamlining customer experiences. Understanding what your business needs, wants and can afford, and then investigating the best options to achieve this, is the foundation for success.
The best starting point is to gain a clear understanding of your current environment and to ask the fundamental question: What do we want to achieve from AI deployment and digitisation in general? The answer will help guide your decision-making and define the way forward, says Dherman. “Once you have a good understanding of your current situation, you will be in a better position to define a clear strategy and implement a step-by-step plan of action that will support your organisation’s transformation journey.”
With AI in place, interactions with customers will move from straightforward transactional models to multidimensional conversations spanning a variety of complementary channels.Kevin Dherman, SYSPRO
Sartini believes AI can be a mammoth task to implement and requires the right foundation for it to be successful. The key to implementing successful AI solutions lies in the company data. “A key challenge to avoid in implementing AI is data silos. For AI to be successful, data needs to be combined to prevent duplication and avoid data drift, which occurs from using outdated or inaccurate data. Getting through this data is the biggest challenge to putting the right foundation in place for AI to succeed. As such, companies need to start with a data refinery that collects and sorts through all the data before implementing an AI solution. Another challenge companies face is opting for AI initiatives that offer the greatest potential for the business without considering the requirements to get to the big prize. AI requires slow and steady progression for it to be successful and companies need to meet several technology and regulatory requirements before they can fully implement AI solutions. As such, companies should adopt a phased approach to AI, starting with automation and moving up towards AI solutions focusing on the short-term gains that each solution offers in the journey towards AI.
AI is really only as effective as the communication channel it’s deployed on, so this needs to be considered too, says Van Wyk. Businesses must look at how they can enable AI on the channels they are already using, for example, SMS, voice or email. If innovation and two-way conversations are required, then businesses need to look at how they will gear to migrate to an alternate channel. A major pain point when it comes to AI is data. AI is about an answer and response mechanism, which is built on a foundation of data. In order to provide appropriate responses and a useful two-way communication channel via AI, the underlying data needs to be up to the job, and built to enable interpretation so that AI can provide a meaningful response. Nothing is more frustrating to customers than a bot that is unable to do so, and poor customer experience will result in low adoption rates, which minimise the return on investment.
It should enable people at different levels in the organisation to engage with AI and receive various levels of data intelligence without them needing to be data scientists. Unless you understand and can extract value from the data, AI will be meaningless. Traditional report writers and data analysts will need to be upskilled in the art of AI, as this is a new paradigm of thinking. While current reports tell you about the business as it is, AI will tell you what the business should be, in order to be profitable, and productive, Dherman adds.
Speaking of the possible pitfalls of AI, Van Wyk says without an effective AI system, a business will frustrate its customers. “If you haven’t planned the AI experience well and executed it properly, your investment will be wasted and adoption rates will be poor. It’s essential to always have the option to transfer to a human being. It’s important to cherry-pick processes to migrate to AI as it is suited to tasks that are particularly predictable, mundane and repetitive. Once an AI system is implemented, it needs to be maintained, updated and continuously optimised. It’s also prudent not to implement AI if it’s not necessary or will not deliver any benefits. Implementing any technology for technology’s sake will inevitably result in unnecessary challenges. Talk to your customers, find out if they would actually use an AI system, and make an informed decision as to whether this is something your business needs or not. As the saying goes, if it ain’t broke, don’t fix it.”
For Sartini, one of the dangers of implementing AI is the potential emotional impact on staff and customers. “It’s a well-established result in social psychology that when people feel anxious, they seek advice from others. Increasingly, AI is being implemented in high-anxiety settings (such as financial services, healthcare, and education).
“Giving customers the potential to be able to interact ‘with a human’ has been shown to reduce anxiety and increase customer satisfaction and trust, even though the majority of customers will not necessarily exercise this option. Knowing that human contact is readily available is important for anxious customers and should be incorporated into the service design when looking at implementing AI for service delivery.
But we are only scratching the surface of how when augmented with day-to-day analytics - AI can help to fuel and lead massive productivity gains in business.Clinton Scott, Techsoft International
Dherman believes that ethics needs to be a consideration, too. “In fact, some researchers believe that AI should hold itself accountable. If AI is used to simply augment human ability, and not replace it, fears around the ‘robot revolution’ can be pacified.”
Of course, no AI conversation can be had without discussing whether it will take away jobs or create new ones. According to Gartner, AI-related job creation will reach two million net-new jobs in 2025. “Although AI will replace some jobs, I believe it’s more likely that AI will augment human ability. In other words, AI will assist humans to do their jobs better. For example, chatbots or ‘digital citizens’ have enabled or augmented human ability by allowing manufacturing businesses to make decisions much faster. What is important is the fact that the chatbot is not replacing the human element in customer service, but rather adding value by offering customers a 24/7 touchpoint. Technology is simply enhancing human abilities in order to place the customer at the centre of a business,” says Dherman.
The beauty of AI is that it doesn’t get tired, it won’t have bad days, and it won’t make decisions based on emotion, ends Noome. “AI will definitely increase productivity because it can work 24/7, 365 days a year. The other side of the coin though is that we should be encouraging human contact and not limiting it any further by technology. My ideal scenario then is that we see humans at the front-end and AI operating in the back-end; humans must keep connecting with each other.”