While many companies are increasingly investing in artificial intelligence to boost productivity, improve customer experience and reduce costs, the gains obtained are still small considering the disruptive potential AI can bring if it is integrated into business strategyDecember 2019 | February 2020
Few technologies appeal to the imaginary as much as artificial intelligence (AI) and machine learning. There are those who believe that in a few years there will be machines so intelligent and autonomous that they will be able to surpass humans in all activities, including those that depend on creativity and decision making, such as writing a poem, running a business, creating a project and designing a car. This may happen, but even in a less fantastic scenario, the disruptive power of AI is immense. This is why a growing number of business leaders have looked to it as an increasingly important tool for business strategy.
According to the study “State of AI in the enterprise”, 63% of the 1,900 executives interviewed by Deloitte said artificial intelligence is very or “critically” relevant to success today. When asked about the importance of AI over the next two years, the percentage rises to 81%. Not surprisingly, they are increasingly investing in this technology: from an average of US$ 3.9 million in 2017 to US$ 4.8 million per company in 2019 – a 23% jump.
Experience and productivity
Technology has been employed for increased productivity, improved customer experience and cost savings. Amazon, for example, uses machine learning to give customer buying recommendations and deep learning (a breakthrough over machine learning) to redesign processes and create new product categories, such as the Alexa virtual assistant, improving customer relationships. Salesforce has integrated its AI-based business intelligence tool (called Einstein) into its CRM software – and delivers about 1 billion forecasts a day to its customers. German retailer Otto uses technology for autonomous operational decision making on an unreachable scale for humans. “Artificial intelligence is already enabling and transforming business models in a variety of industries” says Jefferson Denti, Business Consulting partner and Digital practice leader at Deloitte.
In Brazil, one of the successful cases of using artificial intelligence is Bradesco Artificial Intelligence, a machine learning solution better known as Bradesco’s BIA. Since its inception at the end of 2016, BIA has performed over 100 million interactions with customers and employees, and today has approximately 14 million users.
BIA answers questions about the bank’s products and services via WhatsApp, Google Assistant and Bradesco applications, and helps with operations such as balance inquiries, transfers and loans – in total being responsible for 85 products and services. With a response time of 3 seconds, it can handle 95% of calls. A question like “BIA, how is the economy?” has as the answer up-to-date information about the economic scenario from the bank’s economists.
The next AI frontier
According to Fabricio Lira, IBM Cloud Brazil’s Analytics leader, the use of AI occurs on four fronts: interaction with people, understanding (of objects, images or human emotions), discovery (via data analysis) and, finally, decision making. “Artificial intelligence is used in both the front office and the back office of companies, and the speed with which advances occur makes it difficult to analyze the full potential of the technology.”
The development of artificial intelligence can be divided into three stages. The first is called assisted intelligence. In it, machines are used for collecting and storing large amounts of data, which aid in decision making. The second is the phase of augmented intelligence, in which machine learning acts to expand human analytical skills. This is the stage of many organizations.
The big leap will come when they migrate to the third stage – autonomous intelligence. At this stage the machines are expected to operate in a self-sufficient and effectively intelligent manner. This is the stage in which productivity increases will occur exponentially, as AI can make certain decisions based on an immense amount of data – which no human brain can process.
The initial two phases can be called “narrow AI”. The third is “broad AI”. “At narrow AI, the robot is trained within a domain”, explains Lira from IBM. “At broad AI, the robot has the ability to train itself without limiting itself to one area of knowledge.” For example, it is possible to enable the machine within a reality (or culture) and it begins to interact naturally, seeking self-training and relating diverse areas of knowledge. At this stage, the disruption may come in many forms.
One of the most sensitive areas for organizations in various industries is regulatory compliance. And part of the blame lies in the human bias, which often precludes neutral and impartial analysis of laws and regulations. Artificial intelligence, being free of subjectivities, opens the possibility of an accurate and literal interpretation of a rule, preventing the company from making mistakes in this area, which are not usually cheap for offenders.
Another area where AI can push boundaries is “mass customization” of products and services. Today, organizations make profits by gaining scale or customizing delivery. AI can do both at the same time: customize products and services for mass consumption.
Overcoming the obstacles
In Brazil, the number of organizations that efficiently employ artificial intelligence is still limited. “There has been an advance from last year” says Denti, from Deloitte. “But many companies are still in the early stages when compared to using other digital technologies such as robotics and data analytics.”
One of the common mistakes is looking at technology as an end – not as a tool integrated with the company’s business model. “Some organizations want to have a chatbot just to say they use artificial intelligence” says Cassiano Maschio, commercial director of Inbenta, an AI-based software company that develops chatbots to interact with humans. “The ideal is to adapt the use of technology to your business. That way the company will be innovative, see productivity increases, cost savings and improvements in customer experience.” When this happens, says Maschio, efficiency and productivity gains can reach 30%.
Not connecting technology to business is a mistake that is part of something bigger: the lack of an organizational culture that can think of artificial intelligence in an integrated way. It depends on a number of factors.
The biggest, and perhaps the most challenging, is having well-trained teams (including leadership) adapted to the new reality that AI imposes. Companies also need to have adequate infrastructure for data collection and analysis – which will be used to train the system. And this data needs to be well protected against fraud and privacy violations. “Making good use of technology goes far beyond having technology” says Denti.
Despite the obstacles, these are challenges that must be faced by organizations. After all, with technological developments, artificial intelligence will soon cease to be just a differentiator to become an indispensable technology in business.
AI can be used in many ways by organizations, but the gains occur at a greater scale if it is used within the company's business strategy Jefferson Denti, Business Consulting partner and Digital practice leader at Deloitte
New intelligence in the financial sector
Interview with Bob Contri, Deloitte Global Leader for Financial Services
One area that is beginning to see significant impacts from artificial intelligence is the financial sector. According to the report “The new physics in financial services: Artificial intelligence transforms the finance ecosystem”, produced by Deloitte in partnership with the World Economic Forum, among these changes is the offer of new services, such as autonomous financing, fleeing the “war for the lowest price”. In an interview with Mundo Corporativo, Bob Contri, Deloitte’s global leader for the Financial Services industry, talks about these and other transformations.
What disruptions can the adoption of artificial intelligence cause in financial institutions?
Artificial intelligence has the potential to disrupt customer relationships. By acting, for example, as a counselor, AI allows customers to gain autonomy. Another disruptive area is back office, where operations will be more automated, making room for institutions to focus on core business.
What benefits can artificial intelligence bring to financial institutions?
Several, such as increased productivity and greater fraud detection efficiency. It also helps to minimize risks and overcome regulatory obstacles with transparency. In the artificial intelligence world, the importance of data strategies and partnerships will have a significant impact on financial institutions. In an ecosystem where everyone competes for data diversity, managing partnerships with competitors or potential competitors will be critical, but fraught with risk.
And for the customers?
AI can increase customer engagement, help them make more accurate financial decisions, and improve their experiences. It is a technology that drives the differentiation of institutions in relation to competitors, by allowing greater customization of services.
What changes will occur in the market structure?
Artificial intelligence should favor market extremes. On the one hand, it will open opportunities for smaller, innovation-focused companies that can create services to target specific nests; On the other hand, it will favor large institutions, which are able to achieve large gains in scale, since with AI many operations will be become commodities. In this polarization, midsize firms tend to be harmed. One possible scenario is that, with lower profits, they may be absorbed by the industry giants.
What obstacles exist to the adoption of AI by financial institutions?
The biggest constraint on AI adoption speed is human resources, that is, in adapting talent to a scenario where people will work with the aid of artificial intelligence.
Concerns about data protection and privacy also slow this process. In fact, this is a fear that already negatively impacts AI adoption.