As anyone following AI in 2025 can see, there is (still) a lot of hype. Marketing and expectations have quickly outpaced the success of real-world applications of the technology. The amazing improvement of Large Language Models (LLMs) with the release of ChatGPT 4 in 2023 led many people to believe that artificial general intelligence (AGI) was just around the corner. Self-improving AI, where an AI system enhances its own capabilities and intelligence without direct human intervention, was being touted based on the recent success of LLMs. 

Déjà  Vu

Before the internet changed the course of global business, there was the dotcom boom and crash, which taught us that there could also be a deflation of the AI hype balloon. 

At FastrackPR, we view these new technologies via a host of predictive tools, including the Gartner Hype Cycle. Gartner posits that a potential technology breakthrough kicks things off, is then followed by early success stories that start the hype, but as projects fail to deliver, a trough of disillusionment ensues. This is followed by further learning and then practical applications of the technology (referred to as the slope of enlightenment and plateau of productivity). We’ve been anticipating the deflation of the hype and movement towards “enlightenment”. 

With the release of ChatGPT5, tech leaders, including Sam Altman and Eric Schmidt are lowering AI expectations after being early cheerleaders and investors in the technology. A recent MIT report indicates that approximately 95% of company generative AI pilot programs fail to deliver measurable value.

We view this less as an issue with the underlying technology and more as a learning gap on how to apply the technology to improve organizational workflows (the slope of enlightenment). When humans spend more time double-checking AI outputs than doing the work the AI is supposed to be assisting with, projects stall, imposing an overwhelming "verification tax".  MIT found that successful pilots are characterized by iterative, continuous learning, quantifiable uncertainty, and integration with existing workflows. In our experience, an iterative product approach is the best path forward.

Over the course of our careers, we’ve worked with large institutions to leverage AI as a content quality and search tool. Our practical experiences inform our offerings when guiding clients who are navigating AI. Over the last several years, we have:

• liberated siloed data
• improved workflows, and
• removed some of the drudgery that is better handled by AI.

Ready to see how best to navigate the AI-terrain? Let us help you move up the “slope of enlightenment” to the “plateau of productivity” by applying current AI tools with real value to your project.