Applied Artificial Intelligence: An Introduction For Business Leaders

by Mariya Yao

However, a 2017 McKinsey report found that 41 percent of businesses reported that the uncertain return on investments was one of the biggest barriers preventing them from adopting AI. Because the scope, ongoing maintenance costs, and maturity level of AI technologies vary widely, it is difficult to produce a generalized methodology to quantify ROI for AI. An alternate measure may be to examine how AI technology unlocks business value. Typical metrics center around tangibles such as increased revenue and decreased costs, as well as intangibles such as culture, brand value, and work-life balance.

First, you can assess its revenue creation ability, most notably in external-facing departments such as sales, marketing, and customer service. AI can identify new potential customers for the sales team, facilitate personalization to improve conversion rates and decrease churn, and power customer service bots to provide higher quality service and generate repeat business. Read Chapters 17, 18, and 19 for more detailed analysis on popular AI applications for these business functions.

AI may also allow you to offer new products or services that weren’t previously possible.

AI to offer cisco prducts and services that users were no previously aware they were eligible for.

Measuring the ability to reduce costs is another popular way to assess returns on AI investments. AI promises greater operational efficiencies, predominantly in middle and back office functions, such as in legal, finance and accounting, operations, and human resources. However, efficiency alone is not valuable. Focus instead on the increased output or decreased human capital costs that are made possible by efficiency gains. Don’t forget to include potential cost reductions that result from improved compliance and decreased legal risks. We cover additional details about how AI is used to improve internal business operations in Chapters 13, 14, and 15.

To calculate potential cost reductions, map out the current situation. Follow the operational procedure and note all of the steps that employees must perform as well as the number of employees needed for each task. How long do they need to finish each step? How many total man-hours are spent on the project? What is the fully-loaded cost of these full time equivalents (FTE)? How often is this process repeated?

Portfolio Approach Consider taking a portfolio or a venture capitalist approach to evaluating returns on AI projects.(76) View these early investments as research and development (R&D) ventures and assume that lots of failures will accompany each success. Many of the projects with the biggest ROI also take the longest to mature, require the most investment, and involve the most risk. Therefore, select a variety of projects based on their respective investment requirements, expected time to results, and likelihood of success. Schedule the experiments across multiple quarters and intersperse sure wins with riskier projects. If you’re restricted to testing one project at a time, prioritize likely wins in order to gain credibility and flexibility for future projects.

Pick the Right “True North” Metric How can you tell if your AI strategy is creating long-term and sustainable value? Answering this question can be the most difficult and most impactful step of your machine learning process. Even after you've ascertained that an AI initiative is likely to positively impact the bottom line of your business, you need to define a more specific "true north" metric for each major project in order to keep your machine learning projects on the right trajectory. Goals like "Increase revenue by X dollars" or "Cut costs by Y percent" are almost always too high-level to be useful, since a confluence of intermediate factors drive your end result. Choosing a more specific metric helps you to qualify your AI strategy and also check that your decisions align and advance your business and technology in the right direction. Facebook's true north metric for platform growth is the number of members who connect with ten friends in seven days.(77) Just having a user sign up for an account is insufficient to inspire the engagement rates that Facebook needs to later monetize that user through advertising. Similarly, Slack focuses on teams that have exchanged at least 2,000 messages.(78) Once a team has reached this threshold of usage, they're much more likely to stick around and eventually upgrade to paid plans. Identifying the right true north metric can be a challenge. Ask yourself the following questions to avoid the common mistakes executives make. Is this a...

Business leaders who want to lead AI initiatives at their companies should develop a high-level understanding of how machine learning models are built, even if you are not responsible for writing the code yourself. Your willingness to educate yourself on general technical details will improve your credibility and communication with the engineers on your team.

AI Is Not a Silver Bullet Machine learning is a powerful tool, but it is not right for everything. As with any technology, some use cases will benefit more than others from the application of AI. Each algorithm has distinct advantages that make it more successful in some scenarios but not in others. AI experts and engineers are well-versed in these details, but most executives who lack technical backgrounds tend to clump all AI technologies together and regard it as a silver bullet.

Bad data invariably leads to bad results. Even if you’ve taken care to eliminate errors from your training data, you can still make mistakes with your algorithms. The ultimate goal of any predictive model is to make accurate predictions about unseen data. A good model should first extrapolate patterns from your training data to correctly predict outcomes with reasonable accuracy, then, it should be able to generalize, applying what it has learned to make reasonably accurate predictions on new data.

Science experiments fail 99 percent of the time. As an executive, you’d be fired if you failed 99 percent of the time. Corporate risk aversion prevents most business leaders from leading bold experiments. However, early AI investments can be classified as R&D spending, and they should be regarded as innovation opportunities with potentially exponential returns.

Consistent testing and iteration are critical to AI systems that learn and improve over time. The vast majority of tests will fail initially. When IBM Watson collaborated with Toyota to create advertising copy for the new Mirai car model, the first versions of algorithmically generated texts were incoherent. After a couple months of tuning, the AI system learned to write thousands of useful ads.

Machine learning debt can be divided into three main types: code debt, data debt, and math debt.(86) Code debt arises from the need to revisit and repurpose older code that may no longer suit the project. Data debt focuses on the data that was used to train the algorithm, which may have been incorrect or is no longer relevant. For example, a model predicting consumer healthcare coverage may flounder when healthcare regulations change, as new regulatory mandates decrease the importance of historical data, or when historical data must be purged for reasons of compliance. Finally, math debt stems from the complexity of the model’s algorithms. Most machine learning algorithms will require ongoing customizations that can make them harder to configure, maintain, and understand.

Companies generate revenue by either cutting costs or finding new ways to make money, with the first being generally more straightforward than the second. Current AI-based solutions are very good at reducing inefficiencies in the workplace. By handing off repetitive tasks to software, employees have more time and energy to spend on high-value tasks.

13. General and Administrative General and administrative units such as Finance, Legal, and Business Operations are often underappreciated because they do not generate revenue. However, these functions perform some of the most critical jobs within the company, such as keeping track of the money that Sales bring in, using that money to pay for the ads that Marketing will use to attract new leads, and keeping a wary eye out for regulatory and legal hurdles that Product Management may have to address. General and administrative roles are riddled with tedious but critical tasks such as manual data entry, which requires extreme precision. The exponentially-increasing volume of data combined with limits on the human capacity for sustained attention to detail is a recipe for corporate disaster. Fortunately, computers do not tire and excel at repetitive tasks where attention to detail is essential. Finance and Accounting Instead of forcing your human employees to spend hours over a spreadsheet to audit financial line items for duplicates, errors, expense abuses, and spending anomalies, you can use natural language processing software such as AppZen to automate expense management for accountants and controllers. This cuts down on the possibility of error and removes the most tedious parts of an accounting professional’s job. For employees who must regularly submit expense reports, Expensify saves time and cuts down on error by streamlining business travel booking and expense report...

Current applications include drafting and reviewing contracts, mining documents in discovery and due diligence, sifting data to predict outcomes, and answering routine questions. In-house legal departments can spend 50 percent of their time reviewing contracts, creating bottlenecks that slow down business transactions. AI solutions using natural language processing (NLP) can act as spotter to provide needed information on contract terms, allowing lawyers to focus their attention on the most critical segments of each contract and shortening the overall legal clearance process. Legal teams must ensure that contracts are not hiding obligations, liabilities, or legal exposures. In addition, legal teams must also stay up-to-date with ever-changing compliance rules. AI-powered due diligence agents are already being used to review billions of dollars in mergers and acquisitions (M&A) transactions as well as to extract and manage data of multinational corporations in multiple languages. Legal research can now be supplemented by AI-powered assistants that can review and categorize large bodies of documents, flagging the most important ones for human attention. New products, including ones making use of IBM Watson, can now respond to legal questions in plain English, even outperforming experienced legal professionals in accuracy, efficiency, and user satisfaction.

Your company’s intellectual property is often your most valuable asset. AI can now assist with invention disclosures, docketing, deadlines, filing applications, valuing your IP portfolio, and budgeting. According to lawyers for IBM, the use of AI has cut down on the total time needed to analyze trademark search results by 50 percent. Other uses include spotting warning signs of burgeoning legal issues, which may ultimately help to preserve and increase the value of your brand portfolio.

Most companies have tons of repetitive digital workflows. These workflows can be tedious to complete. Employees responsible for these tasks can easily become bored and inattentive, allowing errors to creep into your operations and your data. Fortunately, these tasks are well-suited for automation by Robotic Process Automation (RPA), which are software robots programmed to perform a specified sequence of actions. Even better, RPA deployment is relatively fast and low risk, so that problematic robots can quickly be removed without detriment to existing systems. Examples of workflows at which RPAs excel include performing regular diagnostics of your software or hardware, creating and updating accounting records (such as payroll), or automatically generating and delivering periodic reports to the relevant stakeholders.

Like customer retention, hiring can be a complicated process. HR must match an open position with the right person who has the required skills. Once hired, HR and the employee then face the new challenge of mapping out a career plan and trajectory that is desirable to both parties. This process may become more challenging if an employee expresses a desire to develop in areas where the company has no open positions. An employee may become dissatisfied and look for jobs elsewhere if that person’s skill set was badly matched with the demands of the job or finds no opportunities forthcoming. Companies like the HR ERP giant Workday and emerging player HiQ apply data science and machine learning methods to understand risk factors and improve retention.

To create meaning, BI must first convert data into information, then analyze that information to create insights that can then be converted into recommendations for action.

As a result of the need to manage data that is increasing in scope and complexity and being generated by multiple business units across an organization, new jobs specializing in the care and feeding of shared data have appeared. Chief Data Officer (CDOs) and Chief Data Scientist positions are now becoming common in companies, especially those interested in championing new AI investments. For companies that do not have the capacity or the desire to tackle data silos on their own, companies like Maana, Alation, and Tamr offer ML-powered data unification and cataloguing services.

Unlike the previous units that primarily generate value through cost-cutting, both Marketing and Sales are directly responsible for generating revenue. This unique position gives the two functions significantly more power to direct investment into new projects. The need to keep up with competitors in fighting for market share means that both units are, on average, much more willing to try new tools as well.

However, AI-based content generation, also known as natural language generation (NLG), has been an area of major development in the last few years. We described how IBM Watson was used to create new advertising copy for Toyota in a previous chapter. Machine learning algorithms can now use content from previous marketing campaigns to create emotional profiles for user groups. Based on these profiles, your NLG solution can create similar content that is tailored to specific platforms and to individual user groups, increasing the likelihood that the targeted audience will engage with your ads.