Navigating the AI Landscape

Thanks to the launch of ChatGPT, the fascination with generative AI is off the charts.? There seems to be no end to the wide range of use cases or doomsday scenarios people can dream up for it. But amidst the hype, it’s important to recognize that there is more than one kind of AI, and what’s good for one business or scenario may be totally wrong for another. There are many types of AI that business leaders can use right now to their advantage. Once you understand what those are and how they can be helpful, you’ll be better equipped to determine the type of AI to leverage and for what in your organization.

It doesn’t take a brain surgeon to understand AI

A useful metaphor to understand AI is to think of ‘left-brain’ versus ‘right-brain.’ Traditionally, the ‘left-hand brain’ is associated with analytical, mathematical, and logical skills. The ‘right-hand brain’, meanwhile, is associated with intuition and creativity.?

To extend the analogy to AI, you could say ‘left-brain’ AI is mostly concerned with making optimal, rational decisions. It needs to sense and interpret all forms of structured and unstructured data to make predictions. It then needs to translate that data into decisions, take a resulting action, and learn from the results. Examples include AI systems that can make intelligent decisions during complex processes such as deciding what offer to make to a customer or how to process an insurance claim to maximize efficiency and effectiveness.

Generative AI sits on the other side – as more of a ‘right-brain’ or ‘creative’ AI that doesn’t necessarily predict or decide anything. Rather, it generates and creates all kinds of useful content from short user prompts to instantly create text, audio, and visual answers to questions, or even create full-fledged prototype applications. This is of course the AI powering ChatGPT, Midjourney, DALL-E2, and all the other content-generating AI that has upended businesses since last fall.

What drives AI?

Simply put, two key capabilities drive both left brain and right brain AI: logical rules-based reasoning and machine learning. I’m an avid chess player, so let’s take that as an example. The first computer chess programs in the 1950s used logic to guide the selection of which moves to evaluate and organized them into something called a Game Tree.? The tree organized the moves and responses of both sides.? Rules were used to focus the selection of moves that might be best in a given type of position, and then the tree was evaluated to determine which path would be the best to follow. ??These trees could get quite complex, sometimes looking ahead a dozen moves or more. These early era games were not able to ‘learn’ from outcomes -- their improvement came through human driven improvements to their rules. ??

Of course, we apply tons of logic in business as well as chess. For instance, all kinds of eligibility, affordability, and compliance rules exist to help decide who gets a loan. Rules-based systems have been part of AI from the start, so it would be a mistake to not see these as AI systems because they are not so-called ‘intelligent.’ ?In addition, there are good reasons to believe rule-based AI will become more important again with the increasing interest in AI regulation and scrutiny around AI ethics.

That said, how can you know whether the logic used is correct or optimal? And how can you improve? This is where machine learning comes into play. Machine learning enables computers to learn from experience – imitating the way humans solve problems and make decisions without rigid and explicated rules that dictate all their actions. These AI models learn the optimal way to act over time through experience. We feed training data into machine learning models so they can start to recognize all kinds of resulting patterns and learn the most optimal moves to make. Machine learning is also the engine that drives generative AI.? For example, large language models learn what text to expect based on text around it, and this can be used to generate new texts or answers to questions. And these systems can be very effective when they are used in combination with more traditional rule-based AI.

Using the same principles, a bank, for example, can use rule-based AI and machine learning to triage the thousands of customer service emails it may receive on a given day and then determine what action is needed. It starts with machine learning models using natural language processing (NLP) to understand what the email is about and predict the customer’s needs – does the customer need a loan, or have an issue with their account?

Rules-based AI comes into play by taking business policies into account to decide on the next best actions – such as which customer based on credit history and other factors is eligible for what type of loan. Machine learning then comes back into the picture after the actions have been taken and outcomes recorded to self-optimize and learn.

What kind of AI is right for you?

So, which form of AI is right for you? ‘Left-brain’ decisioning? ‘Right-brain’ generative AI? Or a combination of the two? The answer depends on your goals. When implemented correctly, AI can play a critical role in driving better customer interactions, improving business processes, and reducing costs. The right application of AI can absolutely improve your mission-critical work.

So, start with your mission and the outcomes you want to achieve. Are you looking to improve customer service, optimize operations, or improve decision making? Think about the availability, quality, and relevance of your data, the systems you have in place, and the skillset or skill gaps of your team. Then choose the AI that best fits your needs right now and test your plan with a small, pilot project and specific use case.

But don’t wait. AI continues to change the way organizations operate, and generative AI specifically is rapidly and radically changing how work will get done. The most important decision to make now is what’s the right AI path or paths to drive your goals.