The Hidden Value of Existing Solutions and the Future of Generative AI

The Hidden Value of Existing Solutions and the Future of Generative AI

Introduction:
In our quest for progress and innovation, we often overlook the power of solutions that have already proven their worth. We underestimate the value of existing best practices and tend to gravitate towards novel ideas. However, in many cases, the answers we seek are right in front of us. This article explores the concept of "doing more of what already works" and applies it to two distinct realms: healthcare and generative AI.

1. The Keystone ICU Project: Unleashing the Power of Old Solutions
   The Keystone ICU Project implemented a checklist strategy in Michigan hospitals, proving that undervalued solutions can have a significant impact on patient outcomes. Often, we underutilize best practices simply because we assume they have already been considered. However, the truth is that "everybody already knows" is not the same as "everybody already does." Progress and improvement can often be found in mastering the fundamentals and leveraging underused insights.

2. The Generative AI Landscape: Uncovering Hidden Value
   When examining the generative AI landscape, we discover that the companies creating the most value have not yet captured the majority of it. Infrastructure vendors, such as cloud platforms and hardware manufacturers, currently dominate the market, while application companies struggle with retention, product differentiation, and gross margins. Model providers, despite being responsible for the existence of this market, have not achieved significant commercial scale.

3. Differentiation and Defensibility: The Key to Success
   To understand the future market structure and long-term value drivers in generative AI, it is crucial to identify the truly differentiated and defensible aspects of the stack. While vertical integration seems tempting, with end-user applications appearing as potential winners, it is not yet clear if this is the only or best path to building a sustainable generative AI business. Margins may improve as competition and efficiency in language models increase. The challenge lies in determining which forms of generative AI products will become standalone companies and which will be absorbed by incumbents.

4. The Importance of Hosting and Commoditization
   Hosting plays a crucial role in the commercialization of generative AI models. Demand for proprietary APIs is growing rapidly, while hosting services for open-source models are emerging as hubs for sharing and integrating models. There is a common belief that AI models will converge in performance over time, leading to commoditization. However, it remains uncertain whether closed-source models can maintain their edge indefinitely.

Actionable Advice:
Before concluding this article, here are three actionable pieces of advice to consider:

1. Embrace the Power of Existing Solutions: Take the time to revisit old solutions and best practices. Often, progress hides behind seemingly boring and underused insights.

2. Focus on Differentiation: When building a generative AI business, prioritize differentiation and defensibility. Vertical integration may not be the only path to success; explore alternative strategies.

3. Keep an Eye on Hosting Services: Hosting plays a critical role in the commercialization of generative AI models. Stay updated on the latest developments and consider the implications for your business.

Conclusion:
In our pursuit of progress, it is essential to recognize the value of existing solutions and best practices. The Keystone ICU Project demonstrates the power of leveraging old insights, while the generative AI landscape highlights the need for differentiation and defensibility. By embracing what already works and exploring new strategies, we can unlock hidden value and drive innovation in various fields, including healthcare and AI.