Why the AI Revolution Is Slowing Down: Institutional Gridlock and a Mediocre Software Ecosystem

The discourse about the speed of artificial intelligence is reaching the realm of existential questions. Many predictions say that AI will bring rapid changes to all aspects of society. But a more grounded perspective shows a different reality: change will be more gradual than expected, and many efforts will benefit from good patience. The reason is not simply technological—it’s rooted in deep institutional fractures, the mediocre quality of our current software ecosystem, and the complex reality of real-world change.

The Firm Institutional Fracture Against Rapid Change

Over the past two decades, we’ve repeatedly experienced promises of rapid disruption that never arrived as quickly as predicted. In 2007, many experts said the U.S. geopolitical position would decline due to peak oil theory. In 2008, financiers believed the dollar system was heading toward collapse. In 2014, many investors were convinced AMD and NVIDIA’s chip manufacturing dominance would end. But each time, deep institutional ties proved stronger than observers expected.

A concrete example is the real estate industry. For decades, real estate brokers earned 5-6% commissions due to asymmetric information between them and consumers. Many believed platforms like Zillow, Redfin, and Opendoor would eliminate the broker’s role. But market inertia and regulatory capture have given real estate professionals a higher survival rate than critics predicted ten years ago.

I experienced this personally. I bought a house a few months ago, and the process required hiring a broker due to numerous regulatory hurdles. Our buyer’s broker earned nearly $50,000 for a transaction involving only document preparation and coordination—tasks that could be completed in just 10 hours. Yet, the market continues to be slow because of institutional friction deeper than software solutions.

Understanding this is not just academic. I built and run a company whose main mission is to help the insurance broker industry shift from traditional “people-driven service” to “software-driven operation.” The most important lesson we learned is a universal law: complex societies take longer to change than calculations suggest. This doesn’t mean no big changes are coming—only that change will be smoother and allow enough time for adaptation.

Software Quality: Mediocrity and Endless Need

Recently, the software sector has experienced price compression due to investor concerns that companies like Monday, Salesforce, and Asana lack protected “moats” in their backend systems that competitors can easily copy. Many analysts believe AI programming signals the end of the SaaS industry because: first, products will become homogeneous and unprofitable; second, many jobs will disappear.

But an important truth is often overlooked: most existing software is genuinely mediocre. I can say this because we spend millions on Salesforce, Monday, and other platforms. Yes, AI helps competitors copy these products, but more critically, AI enables competitors to create better ones.

The problem isn’t change or competition—it’s quality. Every tool I buy is riddled with bugs. Many software applications are so hard to use that I can’t accept their prices—three years ago, I couldn’t transfer money via Citibank’s online banking because of poor system design. Most web apps are not optimized for mobile and desktop use. No product has all the features users need.

Silicon Valley celebrities like Stripe and Linear attract millions of users not because they are perfect, but because they are significantly easier to use than others. Ask a senior software engineer: “Show me a truly perfect software platform,” and the typical response is silence and a look full of irony.

Hidden in this analysis is a deep truth: even at the point of “software singularity,” the need for human labor in software development is practically infinite. We all know the last few percent of improvement always requires the most effort. From this perspective, every software product has the potential to be 100 times more efficient and feature-rich before reaching true saturation.

I believe many commentators claiming the software industry is near collapse lack an intuitive understanding of how software development really works. The industry has been evolving for over 50 years, and despite major technological leaps, it’s always in a state of “not enough.” As a programmer in 2020, my productivity is equivalent to hundreds of developers in the 1970s, and this leverage is truly inspiring. But the clear result is enormous potential for improvement across sectors.

A common misconception here is the Jevons Paradox: when efficiency increases, total demand grows even faster. This doesn’t mean software engineering’s future is permanently secure, but the industry’s capacity to scale labor requirements and institutional inertia surpasses our understanding, making market saturation a very slow process that provides ample time for adaptation.

How Work Can Be Reshaped and Recreated

The transfer of labor and economic power is truly happening—just as autonomous vehicles will disrupt transportation, many white-collar jobs will inevitably change or disappear. For workers parasitic on the system—those who succeed only due to outdated practices and regulatory capture—AI could be a terminal event.

But the U.S. has great hope: the almost unlimited potential and need for industrial transformation and reshoring. You’ve heard of “reshoring manufacturing,” but it’s more than that. We are already losing the ability to produce fundamental components of modern life: batteries, motors, advanced semiconductors—the entire supply chain heavily depends on external sources.

What about military conflict? Or more critically: did you know that 90% of synthesized ammonia worldwide is produced in China? If supply is cut off, no one can produce adequate fertilizers, and the agricultural system collapses. Look at any aspect of physical infrastructure, and you see limitless opportunities for employment—upgrading bridges, roads, power systems. These projects offer tangible validation of work, with bipartisan political support.

We see economic and political trends moving toward this—discussions of manufacturing renaissance, deep technology, and “American power.” My prediction is that when AI truly impacts white-collar professionals, political resistance will target large government stimulus packages for industrial transformation, with “employment-generating projects” facilitating workforce transition.

More compellingly, the physical world has no “singularity”—it’s bounded by friction and complexity. We will overhaul infrastructure—bridges, roads, power grids. People will see more concrete and tangible validation of their work compared to the abstract world of data. A former senior product manager at Salesforce who lost an annual $180,000 compensation might find more meaningful work in projects like the “California Seawater Desalination Project” aiming to end a 25-year drought. It’s not just about building it—it’s about making it excellent and sustainable long-term.

If we are serious about transformation, Jevons Paradox may also apply to the physical world. The unlimited demand for infrastructure and industrial improvements provides a pathway for continuous employment and value creation.

The Path Toward Sustainable Development

The limits of major industrial projects are material resources and wealth. America can achieve self-sufficiency and large-scale production at lower costs. Overcoming resource constraints is critical: in the long run, if AI leaves most white-collar jobs behind, we need mechanisms to maintain high quality of life. Since AI will reduce profit margins toward zero, consumer goods will become dramatically cheaper—prosperity will be automatically achieved.

My assessment is that different sectors will accelerate toward an AI-driven transition at varying speeds, and all change will be more gradual than doomsday scenarios suggest. To clarify: I strongly believe in AI’s potential, and I see a future where cognitive work becomes relatively valueless. But it takes time—and that time gives us the opportunity to make deliberate, well-planned transitions.

At this critical juncture, preventing the crash scenarios others warn of is not impossible. The US government’s response to the pandemic proved their ability to be proactive and aggressive in crisis management. Massive stimulus policies can be quickly deployed if needed. While effectiveness has limitations, this is not the main concern.

The real priority is ensuring material well-being and prosperity for citizens—a comprehensive wealth creation that legitimizes the state and maintains social cohesion, not chasing outdated metrics or economic dogma. If we maintain our intensity and responsiveness to successive technological shifts, we can reach a more secure future without mediocrity or institutional complacency.