AI's Paradox: Making the Easy Harder and the Hard Easier

The AI Productivity Paradox

Artificial intelligence (AI) is revolutionizing the way we work, but its impact is not as straightforward as it seems. While AI makes routine tasks easier, it also increases the complexity of harder tasks, creating a productivity paradox.

A recent discussion at an open forum panel highlighted the themes that engineers face when working with AI. The concerns included sacrificing quality, lack of acknowledgement of current velocity, and the pressure to constantly sprint. The introduction of AI has added another layer to these challenges.

"AI Did It for Me" - A False Sense of Productivity

Developers used to rely on Google, StackOverflow, and GitHub issues to solve problems. Now, they're starting to say "AI did it for me," which can be misleading. If a developer didn't understand the solution provided by AI, it's not truly their work. This raises concerns about the quality and maintainability of the code.

The ease of using AI-generated code can lead to a phenomenon called "vibe coding," where developers rely on AI to write code without fully understanding it. While this might be useful for prototyping or low-stakes projects, it's not suitable for high-stakes environments like healthcare.

• AI assistance can cost more time than it saves, especially when it generates incorrect or incomplete code.
• Using AI as an investigation tool requires a specific skillset that's not easy to master.

The Hard Part Gets Harder

Writing code has always been the easy part of a developer's job. The hard part is investigation, understanding context, validating assumptions, and knowing why a particular approach is the right one. When AI takes over the easy part, developers are left with only the hard work.

Reading and understanding other people's code, including AI-generated code, is much harder than writing code. By offloading the writing task to AI, developers are left with the more challenging task of reviewing and understanding the generated code without the context they'd normally build up by doing the writing themselves.

Sprint Expectations and Burnout

The pressure to deliver results quickly can lead to burnout and decreased quality. When leadership sees a team deliver fast with AI help, they may expect the same level of productivity in the future. However, this can lead to exhaustion and decreased job satisfaction.

The phrase "AI is senior skill, junior trust" highlights the need to treat AI-generated code with caution. While AI can write code like a senior developer, its output should be trusted like that of a junior engineer, requiring careful review and verification.

Ownership and Accountability

Developers need to take responsible ownership of every line of code they ship, including AI-generated code. Cutting and pasting AI output without understanding it can lead to problems down the line, such as difficulties in maintaining or debugging the code.

How AI Can Make the Hard Part Easier

AI can be used to assist with investigation and problem-solving, making the hard part of the job easier. For example, AI can help identify the root cause of a production bug, provide a solution idea, and generate investigation notes.

In one instance, AI helped investigate a timezone display bug by analyzing recent changes and reproducing the issue. Within 15 minutes, the root cause, solution idea, and investigation notes were available, allowing the developer to confirm the fix and deploy it.

The Future of Work

AI is not just changing the nature of work; it's also changing the skills required to succeed. The "average worker" is becoming a thing of the past, as AI takes over routine tasks and increases the demand for workers who can handle complex, high-value tasks.

As AI continues to evolve, it's essential to understand its impact on productivity, job roles, and the skills required to succeed in an AI-driven workplace.