In recent months, the incorporation of artificial intelligence (AI) in business processes has been growing steadily. Most cases are robotization processes and the development of virtual assistants or chatbots, as they are popularly called, with varying degrees of sophistication ranging from simple questions and answers for internal audiences and customers to very sophisticated virtual agents, with object recognition, legacy system integrations and transactional functionality.
Still, the AI experience, when looked at closely, is closely associated with chatbots and simple chat solutions. Some questions arise in light of this scenario: What is the next step in the journey of artificial intelligence? What are the key elements in making my business an artificial intelligence-driven enterprise? Why are new usage cases more difficult to incorporate? How do I make artificial intelligence translate into a competitive advantage and be perceived throughout the enterprise?
Putting AI to Work
Progressive thinking has never been so optimistic. Several studies by research and consulting organizations indicate that most executives see AI as crucial to generating business value in the coming years. IDC estimates that investments will reach $52 billion by 2021. Gartner predicts that by 2020 IA will be one of five investment priorities for more than 30 percent of CIOs. Forrester Research says cognitive computing technologies or IA-based platforms will be worth about $1.23 trillion by 2020. The list of predictions and projections is huge and always points in the same direction.
There are many texts that address a wide variety of recipes for success, but I prefer to focus on the elements that I consider critical to creating effective applications that use artificial intelligence to generate and capture value: use cases, data, talent, culture and ecosystems that translate into various skills that will lead the company to develop new businesses.
a) Use cases
Many companies still invest large amounts of resources (time and money) in the technologies themselves, rather than thinking about commercial problems. Since most companies are still organized in silos, a poorly understood view of the scope and breadth of the technology is normal, leading to initiatives that are limited to the proof of concept that ultimately fails.
The identification of use cases must be aligned with business problems that have not been resolved for some time, or opportunities that visionary enterprises/starting ecosystems do not seize and/or capture. The use of new technologies should enable the digital reinvention necessary to align with customer expectations as well as create and capture value through new business.
IBM estimates that approximately 80 percent of data resides in corporate environments, behind corporate firewalls, so knowing how to organize and exploit this data represents a unique opportunity to generate valuable information and valuable information. Classifying, organizing and democratizing access to this information is fundamental to help in the construction of use cases.
Interestingly, most companies insist on looking for answers to familiar questions, while investing heavily in developing questions to be answered. Market, price, competition and/or external problems are often well known and an innovative answer may be found. The big turning point lies in being able to exploit the large amount of internal data, mostly isolated in silos,
Although most companies delegate responsibility for IT strategy and execution to IT, this strategy must be aligned with the company’s objectives in all units and areas. AI initiatives are best suited to models of collaboration between IT and business areas where experimentation drives the entire process.
Not by chance, the most successful experiences are obtained through agile processes that involve multidisciplinary teams in the construction of a solution. As with all state-of-the-art technology, the training of new professionals requires dedication and investment, since “smart” professionals will rarely be hired easily, hence the enormous relevance in identifying skills and capabilities for talent development.
AI projects involve changes in the way companies operate that significantly affect (or will impact) their way of working and corporate dynamics. As such, a structured change leadership process with strong executive sponsorship is required.
Otherwise, there is a good chance that good initiatives will not be followed for the traditional reasons that affect change processes without effective leadership: group resilience, lack of short-term results, low initial business impact, difficult returns. among others.
It is practically impossible (and increasingly rare) for organizations to innovate and launch new trips on their own. We live in the moment of co-creation, where the ecosystem becomes inseparable from the value chain, adding capabilities that would take a long time to develop individually.
New partnerships and alliances are being formed to enrich content, select specific information or develop capabilities that do not exist individually. A good example is a partnership between IBM and Volkswagen for the development of cognitively capable digital mobility services.
How does it work in practice?
Today’s big challenge revolves around how to develop innovative approaches and develop new business models while maintaining, modernizing, and operating your current systems and models. The question is how to be ambidextrous to execute both perspectives in an integrated, efficient and effective manner. The following is a practical summary of 10 steps to implement an artificial intelligence oriented organization:
Develop a vision of how AI will change the focus of the organization in the next 2 to 5 years, describing one or three objectives to pursue. It is important that this vision is aligned and communicated by the executive team to the entire company. It is this executive support and sponsorship that will enable them to overcome the obstacles and challenges that will arise, ensuring the necessary resilience to move forward;
Identify an unresolved business problem or business opportunity that has been left behind and explore it through the “garage” process that involves Design Thinking and experimentation;
Establish a multidisciplinary team, including the ecosystem, to address the chosen problem rather than the technology itself, avoiding wasting time on infinite proofs of concept. Technology is very relevant, but members with this task will be able to provide the necessary knowledge. The old way of “going home first and then calling a partner to help” has been found to be ineffective in many cases. Leaving co-creation is undoubtedly the best way;
Ensure adequate data governance to guarantee a democratic and reliable information architecture for the entire organization. Data must be catalogued and accessible to those who need it easily and quickly;
Design and validate, with a focus on the end-user/customer experience, the proposed solution to address the chosen use case. Problem analysis is important in specific use cases that can generate ideas related to the main strategy;
Implement the “minimally viable product” or MVP, with the sole objective of validating the solution from a commercial perspective and preparing the team for the next interaction. Good MVPs involve well-defined business objectives, an adopted solution reference architecture, data requirements, security and privacy elements, data topology, operating model, non-functional requirements (access control, environments to be used, performance, etc.) and architectural decisions;
Review issues related to security, privacy and data rights to be used. Increasing regulation and the critical importance of information management play a relevant role before moving on to the next step;
Evaluate and monitor the results of the implementation and prepare for the expansion and evolution of the initial use case by making the necessary course corrections while maintaining the focus and discipline on communication and transparency. It is important to eliminate bias, correct distortions, and ensure the ethics of AI decisions;
Establish an AI Competency Center that will serve as a “consultancy” to scale best practices to more cases of organizational use. Training people with new skills to spread AI in the business environment is critical. The multiplication of skills and the formation of talents is a key factor for the success of the strategy;
10) Errors & Reviews
Promote meetings to review the mistakes and successes found in the different trips. It is important to emphasize that error is part of the learning process, whether human or machine, so accepting error/failure should be the initial premise of any such initiative.
The topic is full of variables and each step can derive a dozen or hundreds of programs and processes, however, the main challenge is to keep the approach simplified and objective. While we develop the ability to work the paradox of producing results for new models, we must keep the current business running.
Fabrício Lira is an IBM Brazil Data & AI executive.
Source: ComputerWorld Brasil