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.

Rightsizing expectations for AI/ML use in your organization

Forward-looking companies should have a roadmap for their AI and ML utilization. Here's what you need to know.

Little robot

[Images: Adobe Stock / ArtistryAlchemy]

Profile photo of Guy Yehiav

President · SmartSense by Digi

As the hype cycle for artificial intelligence (AI) takes its course, enterprise leaders need to determine how they will continue to proactively leverage the emerging technology at the operational level of their organization. Forward-looking companies should have a roadmap for their AI and machine learning (ML) utilization that outlines their short- and long-term objectives for implementation and optimization. 

ENTERPRISES FALLING SHORT 

Amid the digital ubiquity of our current societal environment, the expectations of both customers and employees have been shaped by the widespread use of AI and ML technology in their personal lives. From voice assistants to smart homes and personal use of generative chat applications, people have become accustomed to interacting with AI on a daily basis. However, when individuals interact with enterprises, either as an employee or a customer, their experiences often fall short of expectations. 

In most enterprises, legacy systems and outdated technologies are still relied upon. This is the root problem causing the struggle to keep up with the pace of innovation and integration required for AI and ML implementation. You can’t build AI on top of antiquated ERP, CRM, and EDI systems or “band-aid” type solutions. When enterprises attempt to do so, they end up with novelty outcomes, like so-called AI being able to recite the company hours or other simple tricks that are devoid of material impact. Without a solid foundation of modern technologies and processes, it becomes nearly impossible to leverage the full potential of these advanced tools. The meager results are followed by major disappointment. 

Many employees are using AI-driven technology without guidance or support from their organization because they are familiar with these tools in their personal lives. This creates a gap between what employees are using and what the enterprise is providing, leading to potential security leaks, scalability issues, and an overall lack of processes to support these technologies. Leaders need to put proper processes in place before implementing and adopting supporting technologies. When the technology comes first—especially if it is adopted by the enterprise in a haphazard, unofficial manner—the people will not be empowered to leverage the technology to its full potential. Individual employees may find some short-term success or personal productivity gains, but the enterprise will not realize the advantages in a sustainable way. 

ADDRESSING EMPLOYEE EXPECTATIONS 

Although the picture I’ve painted about the majority of enterprises is bleak, there is a framework for addressing unmet expectations and making progress in implementing practical uses of AI, GenAI, and ML in your organization. Leaders should look at data analysis within their organization through the lens of several steps—descriptive, diagnostic, predictive, prescriptive—and now the evolution, AI and GenAI. Legacy systems will provide descriptive analytics and potentially a diagnostic tool on how to slice and dice the data at an aggregate level. More modern solutions will make predictions from a set of potential outcomes and prescribe corrective actions in order to optimize the outcomes. 

To successfully move to the evolution of true AI, enterprises need to audit their existing analytics and refactor their code to make the data available within a form that allows for real-time training of ML models. If put into digestible formats, the descriptions, predictions, and prescriptions of the past, along with digitized SOPs, can serve as the initial reference material for the ML model. Then, the model can be rewarded and punished with ongoing real-time data to build intelligence and accuracy with few enough false positives to trust its generative outputs. What will start with more accurate prescriptions will evolve into GenAI that can have full on conversations with employees as they problem solve and continuously improve together. 

Another note for enterprise leaders to consider is the potential for leveraging small language models (SLMs) in niche applications. Many business and operations functions are narrow in focus. The intelligence behind their optimization does not need to be able to reference all the knowledge under the sun. While large language models (LLMs) make the headlines, shrewd enterprises will take advantage of SLMs, which require less investment costs while increasing operational efficiency and margin protection. 

By moving through the stages of descriptive, diagnostic, predictive, prescriptive, and finally AI and even GenAI, organizations gradually build a framework that leverages the full potential of these technologies. The key to successfully leveraging AI, GenAI and ML in an organization lies in rightsizing expectations and taking a structured approach to implementation. By focusing on modernizing legacy systems and empowering employees with the tools they need to perform, companies can bridge the gap between employee expectations and enterprise capabilities.