The AI Diary: The Gorman Paradox — Explanation

Dec 24, 2025

In 1950, physicist Enrico Fermi famously asked, “Where is everybody?” regarding the lack of evidence for extraterrestrial civilizations despite the high probability of their existence. In a recent post, The Gorman Paradox applies this same skepticism to the current state of AI software development.

The paradox highlights a glaring disconnect in the tech industry today:

• The Hype: We are told that AI has commoditized coding, allowing working software to be produced for pennies in minutes. The narrative suggests we are entering an era of “throwaway software” where tools like Cursor can instantly generate bespoke replacements for complex SaaS platforms.
• The Reality: Looking at the actual market, this flood of software is missing. There are no AI-generated equivalents of Spotify, Salesforce, or SAP. There are no LLM-generated games dominating the charts, nor is there a statistically significant spike in new products hitting the app stores.

Just as Fermi looked at the stars and saw silence, the “Gorman Paradox” looks at the software landscape and asks a simple, uncomfortable question about the generative AI revolution: If it’s truly this easy to build anything, why haven’t we seen it yet?

Hypothesis

As a software engineer with experience building SaaS and startups, I’ll share my perspective on explaining the Gorman Paradox. And I think the explanation is straightforward.

The reasons we haven’t seen AI-generated equivalents of Spotify, Salesforce, or SAP might be one of the following:

Reason 1 – The product is not the code itself

Developing a software product is not just about writing code. In fact, this is the easiest part.

Behind the scenes, when building a product, there are many requirements that must be fulfilled simultaneously:

• Solving the right problem at the right moment in time
• Having the right strategy to enter the market and acquire customers
• Having the right team members on board with the right skills and connections

Just a quick reminder: there are many examples of failed startups that claimed to solve the right problem with the right team. They worked for months to build a software solution and yet 95% of them fail.

Reason 2 – Complexity

• Context Collapse: Large-scale applications have millions of lines of code with complex dependencies. Current LLMs have a limited “context window” — they can’t hold the entire architecture of a massive system in their “head” at once. They might write a perfect function that breaks a different part of the system they can’t see.
• Technical Debt Speed-Run: AI often generates code that works but is messy (“spaghetti code”). It struggles with best practices, modularity, and avoiding code duplication. If you let an AI build a whole platform, you often end up with an unmaintainable black box that no human can debug or extend later.
• The 80/20 Rule: AI gets you to 80% completion in minutes, but the last 20%—handling edge cases, specific customer bugs, and performance optimization—takes 80% of the time and requires deep, human understanding of why the system was built that way.

Reason 3 – Truest

Software like SAP or Salesforce isn’t just about features; it’s about trust.

• Security & Compliance: Enterprise clients require SOC2 compliance, GDPR adherence, and rock-solid security. AI-generated code frequently hallucinates security vulnerabilities (like hard-coding API keys or using insecure libraries) because it was trained on the entire internet, which includes a lot of bad code.
• Liability: If an AI-generated banking app deletes a user’s savings due to a hallucinated logic error, who is responsible? Most “serious” startups cannot risk their reputation on probabilistic code generation without heavy human oversight.

Summary

The Gorman Paradox exists because AI currently acts as a force multiplier for engineers, not a replacement for engineering. It allows humans to build faster, but it cannot yet replace the human judgment required to build a business (Reason 1), maintain a complex system (Reason 2), or guarantee security (Reason 3).