Companies across all industries are exploring and implementing artificial intelligence (AI) projects, from big data to robotics, to automate business processes, improve the customer experience, and innovate product development. According to McKinsey, “Adopting AI promises considerable benefits for businesses and economies through its contributions to productivity and growth.” But with that promise come challenges.
Computers and machines do not come into this world with an inherent knowledge or understanding of how things work. Like humans, they need to be taught that a red light means stop and green means go. So how do these machines get the intelligence they need to carry out tasks like driving a car or diagnosing disease?
There are several ways to achieve AI, and for all of them, it’s data. Without quality data, artificial intelligence is an impossible dream. There are two ways that data can be manipulated, either by rules or machine learning, to achieve artificial intelligence, and some best practices to help you choose between the two methods.
Long before artificial intelligence and machine learning (ML) became mainstream terms outside of the high-tech field, developers encoded human knowledge in computer systems as rules that are stored in a knowledge base. These rules define all aspects of a task, typically in the form of “If” statements (“if A, then do B, otherwise if X, then do Y”).
While the number of rules that need to be written depends on the number of actions you want a system to handle (for example, 20 actions means manually writing and coding at least 20 rules), rule-based systems generally require less effort, more cost-effective and less risky since these rules will not change or update on their own. However, rules can limit the capabilities of AI with a rigid intelligence that can only do what it was written to do.
Machine learning systems
While a rule-based system could be considered to have “fixed” intelligence, on the contrast, a machine learning system is adaptive and attempts to simulate human intelligence. There is still a layer of underlying rules, but instead of a human writing a fixed set, the machine has the ability to learn new rules on its own and discard the ones that no longer work.
In practice, there are several ways that a machine can learn, but supervised training, when the machine receives data to train, is generally the first step in a machine learning program. Eventually, the machine will be able to understand, categorize, and perform other tasks with unlabeled data or information to itself.
Where to start with an organization’s AI strategy
The anticipated benefits for AI are high, so the decisions a company makes early in its execution can be critical to success. The bottom line is to adjust your technology choices with the underlying business goals that AI set out to achieve. What problems are you trying to solve or challenges are you trying to face?
The decision to implement a rule-based or machine learning system will have a long-term impact on how a company’s AI program evolves and scales. Here are some of the best practices to consider when evaluating which approach is right for your organization:
When it makes sense to choose a rules-based approach
Fixed results: When there is a small or fixed number of results. For example, there are only two states that an “Add to Cart” button can be in, whether it is pressed or not. While it is possible to use machine learning to detect if a user pressed the button, it would not make sense to apply that kind of method.
Risk of error: The error penalty is too high to risk false positives and therefore only rules should be implemented, which will be 100% accurate.
Failure to plan for machine learning: If those who maintain the system have no knowledge of machine learning and the company has no plans to acquire it in the future.
When to apply machine learning:
Simple rules do not apply: When there is no easy to define way to solve a task using simple rules
Speed of change: When situations, scenarios, and data change faster than the ability to continually write new rules.
Natural Language Processing: Tasks that require an understanding of the language or natural language processing. Since there are an infinite number of ways to say something, it is unrealistic, if not totally impossible, to write rules for the normal language. The natural adaptive intelligence of machine learning is optimized for scale.
AI’s promises are real, but for many organizations, the challenge is where to start. If you fall into this category, start by determining whether a rule-based or ML approach will work best for your organization.