Three Questions about AI That Every Employee Should Be Able to Answer
How does it work?
Team members who aren’t responsible for building an AI system should nonetheless know how it processes information and answers questions. It’s particularly important for people to understand the differences between how they learn and how a machine “learns.”
For example, a human trying to analyze 1 million data points will need to simplify the data in some way in order to make sense of it—perhaps by finding an average or creating a chart. A machine learning algorithm, on the other hand, can use every individual data point when it makes its calculations. They are “trained” to spot patterns using an existing set of data inputs and outputs. Because data is fundamental to a machine’s ability to provide useful answers, a manager should ensure that their team members have some basic data literacy. This means helping people to understand what numbers are telling us and what biases and errors might be hidden within them. Understanding data—the fuel of AI—helps people to understand what AI is good at.
What is it good at?
Machine learning tools excel when they can be trained to solve a problem using vast quantities of reliable data and to give answers within clear parameters that people have defined for them.
expenses software is a perfect example: It has the receipts of its millions of users to learn from, and it uses them to help predict whether a cup of coffee from Starbucks should be categorized as travel, stationery, or entertainment. Learning what machine learning is good at quickly helps someone to see what machine learning is not good at. Problems that are novel, or which lack meaningful data to explain them, remain squarely in the realm of human specialties. Help your employees to understand this difference by showing them tools they already use that are powered by AI either within the organization or outside it (such as social media advertising or streaming service recommendations). These examples will help team members to understand AI’s enormous potential and also its limitations.
What should it never do?
Just because machine learning can solve a problem does not mean it should do so. A machine cannot understand the biases that data reveals, for example, nor the consequences of the advice it gives. There may be some problems that your organization should never ask an AI application to solve. For example, you would not want an algorithm to make the final decision for your company on whom to hire, what to discuss at a board meeting, or how to manage a poorly performing staff member. If employees have thought about proper ethical limitations of AI, they can be important guards against its misuse.
The organizations that will do best in the age of artificial intelligence will be good at finding opportunities for AI to help employees do their day-to-day jobs better and will be able to implement those ideas quickly. They will be clear about where to deploy machine learning and where to avoid it. Alongside their investments in technology, they will remind their teams of the importance of human specialties: supporting colleagues, communicating well, and experimenting with novel ideas. To be ready for pervasive AI, an organization’s whole team will need to be ready too.
What Every Manager Should Know about Machine Learning
Not Just Big Data but Wide Data
The enormous scale of data available to firms can pose several challenges. Of course, big data may require advanced software and hardware to handle and store it. But machine learning is about how the analysis of the data also has to adapt to the size of the data set. This is because big data is not just long, but wide as well.
Consider an online retailer’s database of customers in a spreadsheet. Each customer gets a row, and if there are lots of customers then the data set will be long. However, every variable in the data gets its own column, too, and we can now collect so much data on every customer—purchase history, browser history, mouse clicks, text from reviews—that the data is usually wide as well, to the point where there are even more columns than rows. Most of the tools in machine learning are designed to make better use of wide data.
Predictions, Not Causality
The most common application of machine learning tools is to make predictions. Here are a few examples of prediction problems in a business: Making personalized recommendations for customers. Forecasting long-term customer loyalty. Anticipating the future performance of employees. Rating the credit risk of loan applicants.
These settings share some common features. For one, they are all complex environments, where the right decision might depend on a lot of variables (which means they require “wide” data). They also have some outcome to validate the results of a prediction—like whether someone clicks on a recommended item or whether a customer buys again. Finally, there is an important business decision to be made that requires an accurate prediction.
One important difference from traditional statistics is that you’re not focused on causality in machine learning. That is, you might not need to know what happens when you change the environment. Instead you are focusing on prediction, which means you might only need a model of the environment to make the right decision
This is just like deciding whether to leave the house with an umbrella: We have to predict the weather before we decide whether to bring one. The weather forecast is very helpful, but it is limited; the forecast might not tell you how clouds work or how the umbrella works, and it won’t tell you how to change the weather. The same goes for machine learning: Personalized recommendations are forecasts of people’s preferences, and they are helpful, even if they won’t tell you why people like the things they do or how to change what they like. If you keep these limitations in mind, the value of machine learning will be a lot more obvious.
AI Doesn’t Have to Be too Complicated or Expensive for Your Business
Why Adopting AI Outside of Tech Can Be So Hard
Why isn’t AI widely used outside consumer internet companies? The top challenges facing AI adoption in other industries include:
Small data sets. In a consumer internet company with huge numbers of users, engineers have millions of data points that their AI can learn from. But in other industries, the data set sizes are much smaller. For example, can you build an AI system that learns to detect a defective automotive component after seeing only 50 examples? Or to detect a rare disease after learning from just 100 diagnoses? Techniques built for 50 million data points don’t work when you have only 50 data points.
Cost of customization. Consumer internet companies employ dozens or hundreds of skilled engineers to build and maintain monolithic AI systems that create tremendous value—say, an online ad system that generates more than $1 billion in revenue per year. But in other industries, there are numerous $1 million–$5 million projects, each of which needs a custom AI system. For example, each factory manufacturing a different type of product might require a custom inspection system, and every hospital, with its own way of coding health records, might need its own AI to process its patient data. The aggregate value of these hundreds of thousands of projects is massive, but the economics of an individual project might not support hiring a large, dedicated AI team to build and maintain it. This problem is exacerbated by the ongoing shortage of AI talent.
Gap between proof of concept and production. Even when an AI system works in the lab, a massive amount of engineering is needed to deploy it in production. It is not unusual for teams to celebrate a successful proof of concept, only to realize that they still have another 12–24 months of work before the system can be deployed and maintained.
The vast majority of valuable AI projects have yet to be imagined. And even for projects that teams are already working on, the gap that leads to deployment in production remains to be bridged—indeed, Accenture estimates that 80% to 85% of companies’ AI projects are in the proof-of-concept stage.
Here Are Some Things Companies Can Do Right Now
Instead of merely focusing on the quantity of data you collect, also consider the quality; make sure it clearly illustrates the concepts you need the AI to learn.
Make sure your team considers taking a data-centric approach rather than a software-centric approach. Many AI engineers, including many with strong academic or research backgrounds, were trained to take a software-centric approach; urge them to adopt data-centric techniques as well.
For any AI project that you intend to take to production, be sure to plan the deployment process and provide MLOps tools to support it. For example, even while building a proof-of-concept system, urge the teams to begin developing a longer-term plan for data management, deployment, and AI system monitoring and maintenance.