How to Cut Through the Hype Around Artificial Intelligence

By Abhishek Mittal, Vice President, Data Analytics and Operational Excellence, Wolters Kluwer

Any new technology is sure to breed buzz and artificial intelligence (AI) is no exception. Think back to the days when innovations like the telegraph or radio or iPhone first hit the market. Oftentimes, people get excited—and many get ahead of themselves. While hype around a new technology can drive greater adoption, it can also lead to disappointment should things not go quite as planned or promised.

Gartner dubs this journey—from anticipation to disappointment—the hype cycle. An innovation trigger creates an uptrend that eventually peaks. In the case of AI, the first peak of inflated expectations may have been when IBM’s Deep Blue beat chess champion Garry Kasparov. But it’s ill-advised to discuss AI as a broad singular category defined by a single chess game. Every application is at a different stage in the hype cycle. Google has been hyping its chatbot recently but lost a whopping $100 billion in market cap after the tool made an error during a live demo. ChatGPT is another AI chatbot that is receiving tremendous buzz that many experts have labeled as having inflated expectations.

The question is how to cut through the noise to determine if a particular AI solution is approaching the trough of disillusionment or has already waded through it and is delivering real value. No matter the use case, I recommend approaching artificial intelligence with a healthy dose of skepticism. More specifically, ask the following five questions—which I’ll explain using the context of corporate law departments (CLDs)—to evaluate whether the specific application you’re considering is hype or reality.

  1. Is this something you can do without AI? Artificial intelligence is sometimes touted as a silver bullet for the most challenging problems of an organization. But if your department doesn’t understand how it could arrange people, processes, and technology in place to accomplish a particular task without AI, there’s little chance that adding AI will do the trick. Many CLDs bought hook, line, and sinker into hype about AI for contract lifecycle management (CLM), but didn’t actually have a sufficient understanding of CLM to know how AI could help. Thus, it should be unsurprising that most AI offerings for contract lifecycle management were disappointments.
  2. Is the goal well-defined? Another reason applying AI to contract lifecycle management has largely been a disappointment in the legal space is because the technology promised to boil the ocean. AI is much more likely to be effective when it’s tackling something extremely concrete and well-defined. Put simply, ambiguity is a huge red flag. As opposed to promising to manage the entire lifecycle of a contract, for instance, look for a tech that offers a discreet solution to one piece of the lifecycle – say applying a change in one section of a contract throughout the document. Using AI to power legal bill review, which enforces outside counsel guidelines, is much more straightforward and well-defined, as well.
  3. How much data is available? In addition to being more effective with a well-defined mission, AI is also more effective in a well-structured environment with sufficient data. Without data that is centralized, standardized, and anonymized, it will be difficult for an AI tool to do what it claims to. For contract-related AI, there’s often not a sufficient training set to create a useful tool. With legal bill review, on the other hand, mature tools have hundreds of billions of dollars of legal invoices to work with and learn from. Additionally, that data is all in LEDES format, which is used industry-wide. Having a lot of structured data is a huge green flag to move forward with an AI product.
  4. How established is the company selling the product? There are many benefits to buying AI solutions from more mature companies. To start, there’s a good chance they will have larger repositories of data to train the technology on. Additionally, startups often must make grandiose claims to secure funding and drive adoption. Being the new kid on the block is inherently a desperate position to be in. With a larger company, survival is not as stake. Still, further due diligence is warranted. Talk to existing users to assess the credibility of the company you’re considering—and, again, approach everyone and everything with skepticism!
  5. Is it a top priority for your department? Finally, it’s crucial to ensure the AI you’re considering aligns with the goals of your department and, ideally, the company at large. Have a meeting of the minds across the organization to determine the top three to five priorities. If you fail to define your priorities, then technology salespeople are going to define them for you. As you’re outlining priorities, also outline the easiest, most boring way to solve each one. You may not actually need shiny new technology and should only begin vetting AI solutions for use cases that really demand such novel technology. AI, like anything, can be superfluous if not evaluated and implemented properly.

There are many applications where AI can live up to the hype. But there are just as many where it’s destined to fall short. In truth, the hype cycle around AI—or any piece of technology—is neither good nor bad. It’s simply an inevitable pattern, much like the sun rising and setting each day. It’s indisputable that AI-powered tools are becoming increasingly nuanced and sophisticated as they benefit from the accumulation of data. In the legal context specifically, many departments have overcome their initial hesitancy thanks largely to success stories of early adopters. As demand for AI continues to grow, ask yourself the right questions to cut through the noise and find use cases in which the hype was warranted.