The Pernicious Nephew: AI, LLMs, and the Future of Programming

In this episode of the Art and Logic Minimum Viable Podcast, Carlos Perez and Brett Porter dive into one of the most hotly debated topics in technology today: large language models (LLMs) and their impact on software development.

Brett shares a candid view of the current “fog of war” around AI in programming—what’s hype, what’s real, and where these tools genuinely provide value. Along the way, he explores:

- What LLMs _actually_ are (and why “artificial intelligence” may be the wrong label)
- The tension between autocomplete vs. true collaboration with AI tools
- Hallucinations: bugs, features, or sometimes even surprising insights
- The risks of “vibe coding” and why uncritical trust in AI could cost time and money
- Security concerns, from leaked access tokens to agentic AI in the browser
- How junior developers’ career paths may shift in an AI-driven industry
- The client perspective: why rushing into AI projects can lead to “the superpowered, pernicious nephew” problem
- Moments when AI tools have _actually given developers superpowers_—shaving weeks off work by parsing massive legacy codebases or obscure libraries

The conversation balances skepticism with pragmatism, showing both the pitfalls and real opportunities of working with AI in modern software development.

If you’ve wondered whether AI will replace programmers, make their work easier, or just complicate everything, this episode is for you.

You can find us at https://artandlogic.com/

Music by Andrew Sherbrooke @andrew_sherbrooke