# Like Having an Intern? The Impact of LLMs on Software Engineering Careers -Tom Sellek & Netta Avnoon

## Executive summary

The talk analyzes the profound impact of LLMs on software engineering expertise, moving beyond simple productivity metrics. While studies show high adoption rates and immediate gains (e.g., up to 5x increase in lines of code), the discussion highlights significant risks: a potential degradation of core skills, reduced critical thinking, and an over-reliance that impairs long-term learning ability. Experts are cautioned that while LLMs feel like 'interns'—super eager but requiring constant review—this dependency may create a dangerous gap in the junior engineer's ability to independently evaluate or debug complex code.

## Key takeaways

- High Adoption, Low Trust: Despite LLMs showing positive sentiment and high perceived quality (some respondents believe generated code is better than average), a significant portion of developers report low trust in the tool's output, leading to cautious practices like only 8% merging code without human review. This discrepancy suggests a disconnect between perceived capability and actual confidence.
- Skill Degradation is Quantifiable: Research indicates that LLM use can impair fundamental learning abilities. Studies found that using LLMs for tasks led to a quantifiable impairment in understanding and debugging unfamiliar codebases, suggesting the tool doesn't just set a bad example but actively hinders skill acquisition.
- The 'Intern' Analogy: LLMs are often compared to an inexperienced intern: highly productive, available 24/7, but requiring constant human oversight. The core risk is that junior engineers may not develop the necessary critical judgment skills required to effectively review and correct LLM output.
- Cognitive Surrender: The process of over-relying on AI can lead to 'cognitive surrender,' where users are willing to follow the machine's incorrect path, even when it is statistically far less likely to be correct. This goes beyond typical automation bias.

## Technical details

- LLM Adoption & Productivity: A Google Cloud DevOps research study (DORA) reported a 90% LLM adoption rate and found that developers use LLMs for all aspects of development, including writing tests, debugging, and code reviews. Another source noted open-source projects saw between three to five times more lines of code added in the first month using LLMs.
- Code Quality Degradation: Multiple studies suggest that merging LLM-generated code into a production codebase can lead to degradation in overall code quality and architectural cohesion over time. Metrics like static analysis warnings and code complexity metrics show this decline.
- Neuroscientific Impact: EEG measurements comparing LLM users to non-users showed a reduced engagement in alpha and beta networks—neural circuits responsible for focus, suppressing distractions, and maintaining thought patterns. This suggests a physiological difference related to learning.
- Expertise Models: The talk references two models of expertise development: Bloom's Taxonomy (a hierarchical model) and the Dreyfus Model, which posits that experts are not defined by accumulated knowledge but by embodied, intuitive understanding.

## Practical implications

- For CI/CD pipelines: Implement mandatory human review gates for all LLM-generated code, regardless of perceived quality.
- For Training: Focus junior engineers on developing critical evaluation skills (meta-cognition) rather than just coding speed. Treat LLMs as advanced tools, not replacements for judgment.
- For Code Review: Incorporate specific training modules that teach developers how to identify and correct the subtle architectural flaws or logical gaps introduced by AI code.
- For Process Improvement: Recognize that while LLMs boost short-term productivity, they may hinder long-term skill development if used without structured oversight.

## Topics

Generative AI, Software Engineering, Expertise Development, Cognitive Science, Professional Skills, DORA (DevOps Research and Assessment Team at Google Cloud), Anthropic Researchers' Studies, Wharton School Study (Cognitive Surrender)

Source: https://www.youtube.com/watch?v=QjetA7FyqI0