The AI Productivity Paradox: Faster Coding, Lower Quality
Jordan Andersen warns that AI's ability to accelerate development may simply enable developers to produce unmaintainable code at scale.
Artificial intelligence has fundamentally altered the speed of software engineering, but this acceleration may be creating a qualitative crisis in the industry. While AI tools can generate code in seconds, they do not inherently improve a developer's ability to design successful products or write maintainable systems.
In his essay "AI Can Make You Suck Faster Too," published on Hermit Tech, Jordan Andersen argues that the promised "10x speed" of AI-assisted development is a double-edged sword. Andersen posits that while AI removes traditional technical barriers to entry, it simultaneously allows developers to "suck faster" by enabling the rapid production of flawed or unmaintainable code at a scale that was previously impossible. This creates a dangerous illusion of progress where high-volume output is mistaken for actual engineering advancement.
The Math of Innovation
The discussion is rooted in a mathematical observation by Disesdi Shoshana Cox regarding the expected number of revolutionary tech companies that should emerge if development speed increased tenfold. Andersen notes that AI has finally provided the "silver bullet" to solve the problem of programmers moving slowly, yet this leap in productivity has not resulted in a proportional increase in revolutionary companies—such as the emergence of "three AirBnBs" for every one that previously existed.
The Decoupling of Value
This disparity highlights a critical tension in the modern era: the decoupling of productivity from value. Historically, the bottleneck for software success was often the sheer time and effort required to build a functional prototype. Now that the bottleneck has shifted from "how fast can we build it" to "do we know what to build and how to make it last," the primary benefit of AI may actually be the creation of a larger volume of technical debt.
When the cost of generating code drops to near zero, the risk of shipping mediocre software increases. The ability to ship features faster does not equate to an ability to solve complex problems or architect systems that can survive long-term growth. Consequently, the industry faces a future where the volume of software increases, but the average quality of that software potentially declines.
The Path Forward
As AI continues to integrate into the development workflow, the industry must determine how to balance raw speed with engineering rigor. The central question remains whether the efficiency gains of AI will eventually be harnessed to build better products, or if they will simply accelerate the rate at which developers produce technical debt. For now, the evidence suggests that speed without fundamental skill only leads to faster failure.