What’s the ROI of Generative AI?

I’ve been working in IT for over 25 years, through the dotcom boom, the advent of the smart phone, and the sudden shift to hybrid/remote work early in the pandemic. What’s happening right now with the generative AI revolution is like nothing we have seen before.

It’s hard to believe we’re well over a year into exploring the possibilities since OpenAI first launched ChatGPT to the public, sending shockwaves through boardrooms and disrupting every industry—from manufacturing to healthcare to academia. In fact, OpenAI has reported that 92% of Fortune 500 companies use their platform in some capacity. But so many questions remain.

According to our own research, nine in 10 business leaders believe a wide range of roles will be enhanced by generative AI, meaning nearly everyone and every function can benefit from the technology. This much is validated by further research conducted by Harris Poll on behalf of Insight that shows 74% of business leaders are using generative AI for data analysis, 66% as a task copilot, and 63% to generate written content like reports and presentations.

At this stage, the technology industry is hungry for information. Most companies we work with on deploying new digital solutions are still in pilot mode with generative AI. They are still identifying use cases through testing and learning, and we’re finding that many business leaders are asking the same questions.

It’s all centered on extracting return on investment (ROI), and according to our Harris Poll research, nearly two-thirds (65%) said their department has been tasked with defining this.

Yet because the technology is changing so fast, assessing both present and future ROI can be extremely challenging. Here are some ways business leaders can start understanding what is working in their generative AI adoption process . . . and, perhaps more importantly, what’s not working.

Pinpoint how generative AI can address specific business challenges 

I firmly believe that no matter the business, generative AI adoption and implementation needs to be bolstered by a clear “Why?” What problem are we hoping to solve? What is our rationale for using technology as the best means to address it (meaning, why is a generative AI investment uniquely served to achieve our objectives?)

Despite its vast capabilities—from data analysis and product design to written content and personalized chatbots—generative AI isn’t a one-size-fits-all solution. And it’s most certainly not without flaws. Depending on the business challenge, existing digital tools may solve certain problems more effectively. Or more likely, integrating large language models with other technologies (i.e., more traditional AI models and machine learning), may be required to scale a solution for meaningful use across an organization.

We are routinely approached by clients to help them incorporate advanced analytics into the data in their data warehouse. Generative AI is capable of performing advanced analytics but has a limitation of the amount of data it can handle at once. On the other hand, other data lake technology is much more suited for handling larger volumes of data, making it the ideal choice if you have to process large data sets—and parse the right information to the right users to avoid instances of hallucinations.

In this scenario, you must assess the amount and complexity of data present to decide which technology is best. For smaller and simpler data, generative AI would be ideal. But by understanding its limitations and making the right technology choice, your company can ensure efficient analysis, ultimately helping your organization make better-informed business decisions.

Focus on productivity advancements 

To me, what’s most exciting about generative AI is its ability to enhance productivity by making operations more efficient, oftentimes revamping and reinventing processes.

Most industry leaders are asking themselves questions like:

• How can generative AI help my employees save time on mundane tasks?
• How can generative AI help drive better customer service?
• How can generative AI help us create new offerings that weren’t possible a year ago?

A recent MIT study found that the use of chatbot assistants decreased the time it took workers to complete certain tasks—like writing cover letters, sensitive emails, and cost-benefit analyses—by 40%. We’re experiencing similar results at Insight, too.

One example is through a generative AI-powered database that processes thousands of customer contracts, significantly reducing the need for employees to spend time reading through lengthy contracts. In this case, we calculate the tool’s ROI by assessing the time, energy, and resources the technology saves employees. We’re easily making the case for investing in the tool, so much so that our clients are seeking to adopt their own versions of it.

It’s not a matter of replacing people workers because human overlay is still crucial, no matter the task. That said, this concept of “removing the middle” can be a real game-changer for both employees and customers alike—and a very helpful ROI indicator.

Stay curious and open-minded about the possibilities 

There is a growing notion that there’s demo fatigue, meaning that decision makers can become paralyzed on moving forward with a transformation initiative because the possible use cases are limitless and the technology itself is changing so fast. Which way do you take it if you can go in almost any direction, yet your investment resources are finite?

Generative technology isn’t static. It’s constantly learning and evolving. That’s important to remember particularly because the conversational, intuitive type of chat interface we see in a tool like GPT can be a distraction. To me, what’s more exciting is what is actually powering that chat interface.

By removing the need for manual input and interaction, a GPT tool with its advanced AI capabilities can help automate repetitive tasks and save time and money. It also can be integrated with other technologies to increase efficiency, such as robotic process automation and natural language understanding. Using it in this way helps to streamline business processes and improve the overall efficiency of an organization in ways that can bring a hundred times more value than the surface-level application.

Plus, the interface we see today will undoubtedly differ from the interface of tomorrow. As a result, it will help humans do things they weren’t able to do before. That’s what people should be paying more attention to, and that’s where we can expect to see real ROI emerge.

We are only at the beginning of this journey, and the potential is largely untapped. The fact is that most companies can’t even anticipate the sheer ROI generative AI is going to create. However, rather than seeing this as a hindrance, I encourage business leaders to view this white space as an opportunity for creative discovery and new use cases they never thought possible.

David McCurdy is chief enterprise architect and chief technology officer for Insight Enterprises.