On Being an Engineer in 2026

Rapid advances in large language models and AI agents are reshaping what it means to be a software engineer, shifting much day‑to‑day work from writing and debugging code to orchestrating and supervising automated systems. Commenters argue over how far current tools can really go—from automating “CRUD” and outsourced work to potentially replacing whole teams—and what remains uniquely human, such as domain judgment, accountability, taste, and tacit knowledge. Many expect fewer roles for average “ticket‑fixer” programmers, persistent demand for high‑leverage engineers and safety‑minded overseers, and broader social risks if productivity gains outpace new job creation or governance.

Scope of AI Capabilities (Coding, Intuition, “Consciousness”)

  • Debate over whether agents can ever have intuition or feelings; some argue these are tied to human consciousness and biology, others say emergent behavior means we can’t rule machine consciousness out.
  • Several note that LLMs already behave like “system 1” intuition: they produce answers via opaque internal weights and can’t fully explain their reasoning.
  • Tacit knowledge is framed as a training/feedback problem; RL and rich environments may approximate it.

Impact on Software Engineering Work

  • Many report that LLMs can handle most “pre‑AI senior” coding tasks and a growing share of post‑AI glue work, especially CRUD and boilerplate, test generation, refactoring, ticket grooming.
  • Others strongly disagree, saying talk of full automation feels like hype or “gaslighting,” and that real-world systems, security, legacy code, and parallel/distributed programming still expose frequent model errors.
  • Some engineers feel their role has shifted from building and deeply understanding systems to “agent babysitting,” losing the satisfying parts of the job.
  • Others say good engineers who adopt AI become much more productive and can tackle more ambitious projects.

10x vs Average Engineers

  • Widely shared fear: average “ticket taker” developers are increasingly redundant; high‑agency, product‑minded engineers become “100x.”
  • Counterpoint: productivity is hard to measure; labels like 10x/1x are often social and used to justify pressure and precarity.
  • Some emphasize that bad/‑10x engineers are now able to create even more low‑quality code faster.

Business Models, SaaS, and Outsourcing

  • Speculation that AI will undermine outsourcing and low‑end SaaS, as teams “vibe code” internal replacements.
  • Others argue many tools are cheap enough that building in‑house still isn’t worth the risk; AI mostly shifts the build vs buy cutoff.
  • Concern that “software factories” will produce brittle systems where no one owns design quality.

Human Oversight, Safety, and Accountability

  • Common view: the enduring roles are at the “bread” of the sandwich—deciding what to build and validating whether it’s acceptable.
  • Security and governance are recurring worries: agents can amplify blast radius, introduce new attack surfaces, and generate unmaintainable or dangerous systems.
  • Accountability, legal responsibility, and long‑term observability of production systems are seen as fundamentally human bottlenecks.

Economic and Societal Concerns

  • Strong anxiety about job loss, especially for average developers, middle management, and outsourced labor.
  • Some envision new artisan‑style roles (bespoke software for individuals/small businesses); others doubt their economic viability at scale.
  • Broader worries include inequality, potential “ruling class + AI” dominance, and uncertain paths toward any post‑work or UBI‑like equilibrium.