Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

The real breakthrough of generative AI is the data itself

The question of whether “to gen AI or not to gen AI” is really a question about data.

Employees looking at data

[Images: Gorodenkoff / Adobe Stock]

Profile photo of David Keil

BY David Keil

Generative AI has sent the world reeling. Many are fascinated. Others are skeptical. Plenty are scared. No matter what, all are intrigued. 

Although the latest version of OpenAI’s ChatGPT model has transformed work as we know it, it has raised a few concerns among executive leaders. Are the results explainable and trustworthy? What is the legitimacy of digital risks such as IP theft and cybersecurity? 

A HEALTHY DOSE OF SKEPTICISM

As a serial CEO of high-growth technology companies, I have a healthy dose of skepticism for the latest “shiny object,” but I also recognize the opportunities. While I think it’s important—indeed, mandatory—to acknowledge and address the relevant concerns and risks of adopting generative AI as part of a holistic AI strategy, I believe it’s equally important to weigh the potential benefits for enterprise use cases. (Full disclosure: My company’s platform includes a generative AI feature powered by OpenAI.)

Keep in mind that use cases can vary from enterprises making their own models to those integrating existing models for specific applications. This difference is crucial to consider. As Zack Kass, the former head of go-to-market for OpenAI, mentioned, “If you are trying to build your own model, you should be really sure that you should be building your own model. … Training your own model is going to be a fool’s errand for most companies.”

The difference between those who build models and those who partner with LLM providers really comes down to the quantity and quality of data. As Kass explains, “Increasingly, the prevailing belief in the industry with respect to the model providers is that it is both the amount of data but also the quality of data, and the quality of data should basically be equivalent to exceptionally well human-labeled or expert-level data.”

A WORLD OF POSSIBILITIES

To me, personally, these are exciting times. Why? Because the value of unique large language models and generative AI integrations for enterprises has everything to do with data. The better the data, the better the results are for customized enterprise applications.

So instead of stopping generative AI at the door because it’s intimidating and disruptive, enterprise leaders should consider ways they can welcome these models and finetune them to improve efficiency within their own companies, transform existing products or services, or build next-generation offerings. There could be a heart of invaluable data under these models’ menacing exteriors. 

Leaders can’t be afraid to explore the possibilities. If this positioning sounds familiar, it should. Many leaders had the same mind shift more than a decade ago when digital transformation raised deep concerns about “disruptive innovation,” as advanced by Clayton Christensen. “Disrupt or be disrupted” was the mantra brought on by the rapid rise of digitization.

Here we are again: Do we adopt generative AI to maintain a competitive edge? Or do we embrace business as usual? I think the answer is clear.

TRANSFORMING THE IT DESK

In the market space of my company, end user computing (EUC) solutions, ChatGPT integrations can transform the IT service desk. For instance, when level 1 and level 2 support agents can turn to ChatGPT as a co-pilot, they can more efficiently respond to service requests and even offload tedious tasks to the model.

To me, that’s uncovering efficiency and cost-saving opportunities for large enterprises. A continuous stream of data insights has a significant impact on empowering IT to maintain employee productivity and reduce IT tickets. A Forrester study found this insight can reduce ticket volume by 25%, which totals $2.6 million saved over three years for an enterprise with 30,000 employees.   

It is data that drives this capability for EUC use cases. With concerns that generative AI may run out of data to train on, identifying these data-driven use cases is critical to uncover new troves of data sources. For instance, in a large enterprise IT environment, there are tens of thousands of data points that can be collected from endpoint devices, like laptops, phones, and tablets.

Because my company collects more than 10,000 of these anonymized data points every 15 seconds, and has been doing so for decades, we can jump at the AI opportunity before us. Finetuned models could work alongside humans for data analysis to yield even more insights from very large data sets. 

ADVICE FOR LEADERS

My advice to other business leaders is to turn to data as your guiding light on whether integrating generative AI makes sense for your company. Without data, it’s probably best to move on.

The question of whether “to gen AI or not to gen AI,” which others have asked, is really a question about data. The breakthrough technology before us is not necessarily the foundational models built on complex neural networks. Instead, the real “disruptive innovation” of generative AI is the data itself. Where will you take it from here?


As CEO of Lakeside Software, David Keil has a distinguished record of leading enterprise IT organizations through exceptional growth.