LLMs are eroding my software engineering career and I don't know what to do
This software engineer fears that LLMs are eroding the value of domain expertise and 'taste,' turning skilled professionals into 'off-the-shelf' coders. The post resonates deeply with many in tech, sparking a heated debate on AI's impact on job security, the evolving role of engineers, and the future of the industry. It touches on an existential anxiety about what true engineering value remains when AI can 'prompt' years of hard-won knowledge.
The Lowdown
The author of this reflective piece shares a deep concern that Large Language Models (LLMs) are fundamentally devaluing their software engineering career. Having built expertise over years in complex domains like finance and distributed systems, they now observe these hard-earned skills becoming 'promptable,' reducing their unique value proposition. This shift, they argue, is transforming them from a specialized expert into an 'off-the-shelf' engineer, losing the differentiator that once defined their profession.
- The author identifies three pillars of their engineering value: domain knowledge, debugging intuition/distributed system knowledge, and code quality/software architecture ('taste'). They feel the first two have largely been eroded by LLMs.
- The post highlights a perceived change in hiring practices, with companies now seeking generic 'Software Engineers' rather than roles tied to specific domain expertise.
- They lament that 'nobody cares anymore' about software quality or 'taste,' as managers prioritize output, even if it's 'C-grade.'
- Facing this existential professional crisis, the author ponders whether to transition their hobby, like woodworking, into a new profession, questioning the long-term viability of their current career path.
- The core anxiety is not just about job displacement but about the devaluation of accumulated knowledge and the shift in what constitutes valuable engineering work.
This article encapsulates a growing anxiety among experienced software engineers regarding the rapid advancements of AI, particularly LLMs. It directly confronts the uncomfortable question of what remains of a highly skilled, domain-specific career when artificial intelligence can simulate or even surpass human capabilities in core tasks, leaving many to wonder about their professional future.
The Gossip
The Evolving Role of Engineers
This discussion centers on how LLMs are fundamentally altering the software engineering landscape. Some argue that traditional domain expertise and even 'taste' in code quality are being devalued, making experienced engineers feel like 'off-the-shelf' commodities. Others counter that LLMs, while powerful tools, are still error-prone and require human oversight, especially in complex or critical systems. The debate touches on whether engineers will become 'prompt engineers,' 'software designers,' or whether their roles will shift to higher-level architectural and problem-solving tasks, or simply supervising AI output.
Domain Expertise Debated
A significant point of contention revolves around the enduring value of specialized domain knowledge. The author laments that their deep understanding of finance and payment systems is now 'promptable.' Commenters debate whether domain expertise is a lasting competitive advantage or if it's more quickly acquired than fundamental engineering principles. Some argue that nuanced, business-specific knowledge remains a crucial human differentiator, particularly in highly regulated or complex fields like fintech, where LLMs routinely fail at specific intricacies.
AI's Broader Economic Implications
Beyond individual careers, many commenters explore the larger implications of widespread AI adoption. Concerns range from a structural shift worse than past economic downturns to a dystopian future where only capital owners thrive and human labor is devalued to nothing, leading to societal collapse. Conversely, some express optimism, viewing AI as a liberator from 'forced labor,' allowing humans to pursue hobbies or more creative, 'artistic' coding. The discussion also touches on liability for AI-generated errors and the increasing pressure for 'velocity' in industries like fintech.
Management's Mandate and Liability
Commenters note a clear push from management to integrate LLMs, often prioritizing 'delivery velocity' and cost-cutting by reducing SaaS subscriptions in favor of AI-generated code. This drive to automate everything and 'take humans out of the loop' often comes with a dismissal of engineers' concerns. A critical point raised is the issue of accountability and liability, especially in industries like finance where mistakes can be catastrophic; engineers traditionally provide this ownership, a role that LLMs cannot fill.