Build the AI Business Case A CIO’s guide to building the strategy and business case to implement AI in the enterprise EDITED BY Whit Andrews, Vice President and Distinguished Analyst © 2018 Gartner, Inc. and/or its affiliates. All rights reserved. Gartner is a registered trademark of Gartner, Inc. or its affiliates. For more information, email [email protected] or visit gartner.com. PR_440132
Introduction 2 Artificial intelligence (AI) is still the stuff of the future for some organizations. But strategic integration and experimentation with AI has already led to new revenue, better business models and innovative products. You can create and implement a plan for AI in your business. Embrace the challenge. Experiment and introduce the technology to the board and to your employees. With this e-book, CIOs will learn how to create a business case for AI, how to harness current trends and the importance of using good data to implement AI in the enterprise. Whit Andrews Vice President and Distinguished Analyst, Gartner Build the AI Business Case
Contents CHAPTER 1 CHAPTER 2 CHAPTER 3 04 14 19 Build Your AI Implement Build the Business Case AI Technology Right AI Data Build the AI Business Case 3
CHAPTER 1 Build Your AI Business Case Build the AI Business Case 4
Build Your AI Business Case When a company realized that up to 30% of calls it received For example, the interaction between the company were from customers asking about order status, its leadership and the consumer provides data, such as they order wanted to know if AI would be able to help manage the X amount of Y products every Z weeks. The company interactions. The short answer was yes, a virtual customer can then use AI to further enrich the relationship assistant could answer questions ranging from “Where is beyond that interaction. my order?” to “How long will I have to wait?” But the bigger question was whether AI could help the company in even During future interactions, the data might enable the more impactful ways. seller to ask questions specific to the customer, such as “We know you are frequently waiting on delivery. “L ook at how you are using Would you like to subscribe to this product or order technology today during critical larger quantities?” AI enables companies to collect data interactions with customers — from a wide variety of places and apply self-improving business moments — and consider analysis that can take action — and on a level of how the value of those moments granularity never before available. could be increased.” A recent Gartner survey of global CIOs found that only Whit Andrews 4% of respondents had deployed AI. Gartner estimates Vice President and Distinguished Analyst, Gartner that by 2020, AI will be a priority for more than 30% of CIOs. However, the survey also found that one-fifth of the CIOs are already piloting or planning to pilot AI in the short term. Build the AI Business Case 5
Build Your AI Business Case AI basics Common AI applications Common definitions of AI focus on automation and, as Typically, common AI applications analyze contextual a result, often miss the hidden opportunities available to interaction data combined with historical data in real time. IT and business leaders. AI is technology that emulates human performance, typically by learning from it. Sales and marketing: Customize the sales process, personalize communications to The most common mistake with AI is to focus on prospects and clients, match sales staff automation rather than including augmentation of human to buyers and offer personalized pricing. decision making and interactions. If CIOs focus only on further automation via AI, they miss the hidden opportunities Service: Offer virtual customer assistance for greater personalization and differentiation. AI can augment and triage, predict maintenance and upcoming humans, as it has the ability to classify information and make repair needs, connect service staff to customers predictions faster and at higher volumes than humans can and discover process gaps. accomplish on their own. Supply chain: Discover and correct data errors, CIOs should look for critical business points where human discover risks in the supply chain, elevate insights interaction or human expertise adds value. They then should from Internet of Things (IoT) devices in the field consider how AI might augment those efforts to create even and plan logistics. more value. Banking and financial services: Help customers First, assess which business outcomes would benefit most access their bank balances using chatbots. from AI. Second, evaluate how AI might help achieve those outcomes. Normalize AI planning and development for your Healthcare: Follow up with patients post-discharge entire organization, including leaders of data and analytics, using virtual nursing assistants. applications and lines of business. Build the AI Business Case 6
Trends in AI Build Your AI Business Case Gartner expects that three trends will affect AI in the next Gartner estimates that few years. by 2020, AI will be a priority for more than 1. Better communication (both ways) with people: 30% of CIOs. Natural-language processing, generation and contextual interpretation will make AI easier to use and will improve the use of all computing resources. 2. Deeper and broader integration with existing applications and IoT projects: AI has its greatest value when it is built into architectures that drive business and service value. 3. Richer ecosystem interaction: As AI becomes more common, applications that employ it must work effectively with others employing similar technologies, which will result in chains and meshes of AI systems that work simultaneously toward their individual goals in a cooperative but decoupled fashion. Generally, AI will trend from one-off experimental projects to an approach that integrates the technology with the business. Build the AI Business Case 7
Breaking down the intelligence Build Your AI Business Case AI can range from low-intelligence applications like Learners gather information from multiple sources automation to higher-end intelligence capable of decision to solve complex problems (Watson, wholly making. It can also be controlled centrally or distributed automated military drones). across multiple machines. What makes sense for your company will depend on budget, risk appetite, resources Creators initiate a paradigm shift, such as inventing and other factors. a new business model. They are not merely According to Jorge Lopez, Gartner vice president and tools that people use. They have the potential to distinguished analyst, AI can be broken down into five engineer actions harmful to humans. They will levels of sophistication: change humans’ relationship to technology as well as people’s roles within society and the economy. Reactors follow simple rules but can respond to Therefore, AI creators require profound thought changing circumstances within limits (such as before development. basic drones). These five models have three types of organization: Categorizers recognize types of things and can Stand-alone — The individual entity acts by itself to solve take simple actions to deal with them within a problems. The enterprise exercises centralized control over controlled environment (warehouse robots). it by overseeing the entity as it performs. Responders serve the needs of others by figuring Federation — Multiple versions of an entity work in the same out questions and situations (driverless cars, way but on different problems (e.g., robo advisors, personal personal assistants). assistants). The enterprise can exercise central control or give more autonomy to the entities. Build the AI Business Case Swarm — Multiple entities work together on the same problem (e.g., Intel light show drones, Perdix drones). Control over execution is left to the machines entirely or requires only light human management. 8
Use Cases for AI Application Categories Build Your AI Business Case AI Strategy Framework With Examples Swarm Event lights Reconnaissance Smooth show traffic flow Organization of Individual Entities Federation Faster data Speedy order Expanded High-speed entry fulfillment market trading Stand-Alone Automated Improved Mobility as Medical claims handling customer a service diagnostics service Reactor Categorizer Responder Learner Creator Low High Individual Entity Intelligence Source: Gartner 9 Build the AI Business Case
The case for AI investment Build Your AI Business Case Once you’ve developed a rock-solid understanding of AI and Focus on answering these four questions its potential applications, it’s time to make a case for a pilot. when you want to define an AI project: Despite the enterprise-level interest in AI projects and their potential to fundamentally change the dynamics of business 1 Why are you doing this project? value, most AI technologies are nascent at best. According to 2 For whom are you trying to deliver a recent Gartner survey, 37% of organizations are still looking to define their AI strategies, while 35% are struggling to this solution? identify suitable use cases. 3 What solution and technology 37% of organizations are framework will you employ? still looking to define their AI strategies. 4 How will you deliver this project? To secure financial backing for a pilot, CIOs will need to put forward a use case. This can be particularly challenging for AI, as there is no such thing as an AI business case. Instead, the business case will be for a particular business scenario, problem or use case that employs AI methods and techniques as part of the overall solution. Build the AI Business Case 10
How AI Business Cases Differ From Other Business Cases Build Your AI Business Case To build a successful business case for AI projects, CIOs need to articulate and address the specific factors around how AI projects differ from other IT solutions. 1 AI solutions can seem costly without 3 AI business cases will require substantial providing any immediate gain cultural change AI projects can appear costly without any immediate For most enterprises, the mindset shift required for gains — particularly for loosely bound scenarios and AI can lead to “cultural anxiety” because it calls for in organizations that aren’t used to setting aside a deep change in behaviors and ways of thinking. budget to develop and deploy solutions for new CIOs should acknowledge the cultural change, be business scenarios. proactive in managing related challenges and build trust over time. Cultural change and successful 2 AI projects require different technology transitions to new roles and practices are dependent and problem-solving skills on open dialogue and mutual respect among IT members and between management and staff. Talent acquisition is likely to be one of the biggest barriers to AI adoption going forward. 4 The need to spend more time on data, training and algorithms Although long-term strategies should include how to leverage academic communities and open- AI depends on data and the interactions of source technologies to ease the lack of resources, algorithms, so when it comes to AI business the immediate priority is working out what needs to cases and plans, success is largely dependent happen now. Leverage and train existing resources — on robust data and analytics infrastructure. particularly on data science tools — as a key strategy. Organizations looking to move forward with AI tech must be constantly learning and iterating on algorithms and on how to select, prepare and apply data. Build the AI Business Case 11
How AI Business Cases Differ From Other Business Cases Build Your AI Business Case 5 AI is the representation of a decision 7 AI algorithms carry new model, rather than a process event governance demands The same characteristics that make AI a good AI is a tricky technology because algorithms can solution for dealing with data are also the attributes operate independently of human actors. This that make it difficult to support with a business case. means that they must act on behalf of all actors. Traditional analytic models use fixed rules to arrive at Ensure transparency and openness in all decisions. a conclusion. However, AI analytics use dynamic data Furthermore, ensure that governance discussions and heuristic solutions to arrive at conclusions that happen in parallel with technology discussions. might elude traditional methods. This makes them valuable, but difficult to predict. 6 The critical decision to build, buy or outsource The decision to build, buy or outsource depends on both the project and the company. Ask these questions to decide what might work for the company: How complex is the project? What is the timeline? Is there budget available? How urgent is the need? Use the answers to make your decision. Build the AI Business Case 12
Build Your AI Business Case Explaining AI to the C-suite has to understand both the technology and the potential business impact of it, and be able to relay that in business At a conference on leadership, a CEO of a large wealth terms to executives. management firm attended a session on AI. She learned that other firms were using robo advisors to offer low-cost wealth Address how the technology will enable CEOs to reach their management to clients without enough holdings to justify objectives to save money, grow revenue, increase operational using a traditional wealth management firm. Recognizing efficiency or innovate products. For example, a large mining the potential revenue growth, she asked the CIO to create a company is using AI to control 69 driverless trucks at mine presentation for how AI could help the company. sites. The trucks are controlled from a centralized operations center, offering reduced head count and cost savings. In this At some point, most CIOs will be called to explain AI to example, the focus should be on the savings from reduced the CEO and executive board. personnel, not necessarily how the technology controls the driverless trucks. CEOs need to hear how AI will address “C reate a version of the Gartner AI mission-critical priorities to justify resources. strategy framework, with examples of each category of AI application that are For a presentation, use real-world examples to showcase relevant to the enterprise’s business. It how the technology could affect the business. This will will help you prepare strategic options allow for strategic suggestions related directly to the that you can recommend to the CEO.” business and help shape an action-oriented plan. Offer concrete solutions, instead of theoretical technologies, Jorge Lopez when speaking to the business. Vice President and Distinguished Analyst, Gartner For example, when a global cosmetics firm wanted to The CIO needs to speak in the language the CEO speaks, innovate, the company acquired a startup that enables which is how AI will help the business strategy. The CIO customers to try on makeup virtually and offers personalized options for products to purchase. The acquisition enabled the company to keep up with the modern customer and increase the number of product suggestions. Build the AI Business Case 13
CHAPTER 2 Implement AI Technology Build the AI Business Case 14
Implement AI Technology After six months of planning, Kelly, a CIO at a global finance Learn from other companies company, got the green light from the CEO to explore what AI could do for the company. In her presentation, she explained When initiating an AI project, look to take advantage of that AI could innovate offerings for clients to include options lessons learned by other companies. like automated trading, and chatbots could cut down on the volume of basic questions call centers received. Now, Kelly Be realistic about a timeline needs to bring those ideas to life. Once you have approval from executives, it can be tempting “Gartner predicts that, by 2022, to think a pilot project will follow quickly. Organizations enterprise AI projects with built-in tend to underestimate how long it will take to overcome transparency will be twice as likely to complexities and get AI off the ground. In fact, according receive funding from CIOs.” to results from our 2017 Annual Enterprise Survey, 58% of respondents in companies currently piloting AI projects say it Whit Andrews took two or more years to reach the piloting phase, and only Vice President and Distinguished Analyst, Gartner 28% of respondents reported getting past the planning stage in the first year.1 Build the AI Business Case 15
Expected vs. Actual AI Project Planning Time Implement AI Technology Expected We will plan We will 3 4 AI project and begin launch! duration Still Launch pilot 2 piloting happens! Years 1 How long it We started Still really takes planning planning . Then a pilot Source: Gartner 16 1. 2017 Annual Enterprise Survey Build the AI Business Case
What lessons can we learn from these early Implement AI Technology AI pioneers? Focus on worker augmentation, not worker Aim for fairly “soft” outcomes, such as replacement improvements to processes, customer satisfaction, AI’s potential to reduce staff head count attracts the products or financial benchmarking attention of senior business executives as a potential cost-saving initiative. A more informed expectation, When Gartner Research Circle respondents were however, would be for applications that help and asked what lessons they had learned from early improve human endeavors, as AI promises benefits AI projects, many urged others not to fall into the far beyond automation. Organizations that embrace trap of seeking only immediate monetary gains. this perspective are more likely to find workers eager They advised instead to aim initially for less- to embrace AI. quantifiable benefits from which financial gains would eventually arise. These might come from Plan for the transfer of knowledge from external “softer” or more “open” outcomes, such as service providers and vendors to enterprise IT and improved marketing or brand identity, or they business workers could lead to wider benefits altogether. External service providers can play a key role in planning and delivering AI-powered software. When Some companies need to demonstrate financial using an external service provider, knowledge benefit to initiate an AI project. In such cases, it transference is crucial. Organizations need to plan makes sense to pursue small-scale plans likely to for the transfer of knowledge from external service deliver small-scale payoffs that will offer lessons providers and vendors to enterprise IT and business for larger implementations. workers throughout the execution of projects. This will ensure that the internal skills are available to develop similar capabilities in the future. Build the AI Business Case 17
Choose AI solutions that offer means of tracking Implement AI Technology and revealing AI decisions Own the governance and the skills It’s important to opt for AI solutions that offer means of tracking and revealing AI decisions — ideally Recent breakthroughs in machine learning, big using action audit trails and features that visualize data, computer vision and speech recognition or explain results. To that end, Gartner predicts are increasing the commercial potential of AI. But that by 2022, enterprise AI projects with built-in AI requires new skills and a new way of thinking transparency will be twice as likely to receive about problems. IT must own the strategy and funding from CIOs. governance of AI solutions. Start small; don’t worry about immediate ROI Pilot AI experiments can start with a small investment of skill and resources. Full production CIOs should begin with small experiments that are rollout requires a bigger investment in building purely for learning. Lose the pressure to get an initial and retaining the necessary talent. These skills ROI and use the time to pilot projects that employ include technical knowledge in specific AI a variety of technologies to assess which make the technologies, data science, maintaining quality most sense for the business. data, problem domain expertise and skills to monitor, maintain and govern the environment. Also, expose the business to available AI options. Drop an Alexa, Cortana and a drone in a room and tell employees to play around with the toys. Reach out to software companies to gain access to free cloud demo environments and let employees experiment without the pressure of needing an immediate return. Build the AI Business Case 18
CHAPTER 3 19 Build the Right AI Data Build the AI Business Case
Build the Right AI Data Given the pervasive nature of the technology, the including the purpose or reason that the AI solution is consequences of IT getting AI right or wrong are potentially being developed, as well as other key terms, such as profound. When used incorrectly, AI can unintentionally the types of data used and gathered from the solution. reinforce harmful biases, increase polarization and result in other damaging consequences. Use AI right “With the excitement for and hype The information language barrier can exist locally or surrounding the possibilities of AI, it systemically, regardless of program scope or organizational is easy to focus on the technology and maturity. Addressing it requires a mindset shift, as well as coding disciplines — what might be deliberate acknowledgment and intervention to course- considered the ‘artificial’ aspects.” correct. To make data literacy more explicit, CIOs should develop a data literacy program. Alan D. Duncan Vice President, Gartner Identify fluent and native speakers who speak data naturally and effortlessly. Fluent speakers should be adept Build the case for data literacy as a new core competency for at describing contextualized use cases and outcomes, the both creators and consumers of AI. Follow three steps: Build analytical techniques applied to them and the underlying AI right, use AI right and keep AI right. data sources, entities and key attributes involved. Build AI right Identify skilled translators. Classic translators are often enterprise data or information architects, data scientists, To “build AI right,” it is key to first establish the basic information stewards or related program managers. vocabulary of AI — a technical dialect of how people “speak data.” At the very least, CIOs should determine the primary Identify areas where communication barriers are inhibiting terms used when describing an AI system or solution, the effectiveness of data and analytics initiatives. Pay particular attention to business-IT gaps, data-analytics gaps and veteran-rookie gaps. Build the AI Business Case 20
Build the Right AI Data Actively listen for business outcomes not clearly differentiation and a superior value proposition; articulated in terms of explicit action. What business dangers include reputational risk, regulatory issues moments are being enabled with enhanced data and and financial losses. analytics capabilities? What operational decisions are being improved? Use AI algorithms and data exchange as enablers of digital interactions, and a way to enable stakeholders Identify key stakeholders requiring specialized translations. to participate in an ecosystem rather than as specific To assess data literacy levels, ask key stakeholders to process controls. Encourage everyone contributing their articulate the value of data as a strategic asset in terms data within the AI environment to be active participants of business outcomes, including enhanced business in a mutually beneficial ecosystem. moments, monetization and risk mitigation. Conclusion Identify and maintain a list of words and phrases. Engage the data and analytics team in crafting ways to better AI offers great potential to companies and CIOs who articulate these phrases. implement it well. Keep AI right Focus on three main tasks when developing an AI strategy. First, link the technology with the business strategy and Not even the most successful companies can afford to think desired business outcomes. Second, decide how to they are immune to ethical mishaps. Extensive and explicit implement the technology across the business. Finally, discussion is needed to distinguish between the types of facilitate the transition from planning to implementing ethical questions and dilemmas one can face versus the by researching early AI projects and lessons learned. actual ethical position one can take. Now is the time to experiment and think about how the Take a step back and absorb digital ethics and digital technology could influence your business model of the connectivism as a philosophy for the improvement of future. Remember to think in terms of outcomes, not digital business — and digital society more generally. technology, and articulate the reasons an executive board should invest in AI now versus later. Finally, look to bring Actively look for ethical case studies relating to the use on or seek out specialized skill sets to handle the data of data in AI, as the ethical questions that confront you influx and new technology. are often not new. Opportunities include competitive Build the AI Business Case 21
Additional Research Chapter 1 Chapter 2 Smarter With Gartner articles Smarter With Gartner articles How to Build a Business Case for Artificial Intelligence Lessons From Artificial Intelligence Pioneers Moutusi Sau, April 2018 Whit Andrews, February 2018 The CIO’s Guide to Artificial Intelligence Chapter 3 Whit Andrews, January 2018 Smarter With Gartner articles Client research How to Get Artificial Intelligence ‘Right’ Seven Factors That Make Business Cases for Artificial Alan D. Duncan, April 2018 Intelligence Projects Different Moutusi Sau, Whit Andrews, Alan D. Duncan, February 2018 Client research Artificial Intelligence Demands That CIOs Foster a Data- A Framework for Applying AI in the Enterprise Literate Society Bern Elliot, Whit Andrews, June 2017 Alan D. Duncan, January 2018 Craft an Artificial Intelligence Strategy: A Gartner Trend Notes: Insight Report 2017 Annual Enterprise Study Whit Andrews, January 2018 1. This study was conducted to further understand the current enterprise How CIOs Can Analyze the Strategic Possibilities of AI technology landscape, with this section focusing specifically on the perspective of Jorge Lopez, January 2018 more senior-level employees. The research was conducted online from November 2017 to December 2017 among 1,990 total respondents in organizations with more Guide CEOs Toward an AI Plan Based on the Business Strategy than 20 employees located in the United States, U.K., France, Brazil, China and India. Jorge Lopez, January 2018 Of those, 890 respondents qualified to answer the Artificial Intelligence (AI) section. To qualify for the AI section, respondents were required to be from an organization Develop Your Artificial Intelligence Strategy Expecting These that is currently investing in AI (no description/limitations on investment), report Three Trends to Shape Its Future that they are at least in the planning stage and be personally involved in any of Whit Andrews, April 2017 these decisions: Implementation, planning/budgeting, evaluating vendors or setting strategy. Build the AI Business Case The results of this study are representative of the respondent base and not necessarily the market as a whole. 22
Learn more. Dig deep. Stay ahead. Free content: Visit Smarter With Gartner For more Smarter With Gartner insight, subscribe to our weekly newsletter.