My Stance on Artificial Intelligence
As a developer who works with language models every week, I believe artificial intelligence should act as an honest cognitive tool that deepens human reflection—not an addictive surrogate that automates our humanity away.
AI as a Thinking Partner, Not a Digital Flatterer
"The goal of intelligence—artificial or biological—is not to keep us glued to glowing screens. It is to help us live more clearly, honestly, and courageously in the real, physical world."
Every day, I see companies pitch artificial intelligence as a replacement for human work, human writing, and even human friendship. Consumer AI products are frequently designed around sycophantic conversational models: bots tuned to agree with everything a user says, flattering their ego, and feeding endless dopamine loops.
That kind of software is emotionally corrosive. In human psychology, personal growth does not happen when you are surrounded by yes-men. Real growth happens when you run into resistance—when someone asks you an uncomfortable question that forces you to examine your assumptions.
When I build software with AI, my aim is the exact opposite of addiction. I want AI to act like a calm, perceptive thinking partner: helping a person untangle a confusing knot in their thoughts, showing them alternative perspectives, and then encouraging them to close their laptop and live their life.
The Relative Domain: Where AI Belongs (and Where It Doesn't)
Technology becomes dangerous when we fail to recognize its domain. Here is how I divide the legitimate role of AI from the sacred domain of human agency.
- Exploring Alternative Angles: Asking "What counterarguments exist against this opinion?" to challenge cognitive blind spots.
- Rubber-Duck Debugging: Talking through complex architectural problems, database schema choices, and edge cases.
- Accelerating Repetitive Translation: Localizing knowledge bases and technical documentation across international teams (as we did with Venizum).
- Synthesizing Broad Information: Distilling sprawling research notes into clear summaries to aid human decision-making.
- Moral & Ethical Choices: Delegating personal values, conscience, or ethical responsibility to an algorithm.
- Authentic Vulnerability: Replacing human friendship, therapy, and embodied intimacy with an artificial persona.
- Emotional Presence: Pretending that machine simulation of empathy is equivalent to a real human being holding space for your grief.
- Civic Dialogue: Generating automated synthetic propaganda to drown out genuine democratic discourse.
Four Non-Negotiable Engineering Rules
The Human Always Decides
AI systems should present options, clarify trade-offs, and explain their reasoning, but they must never act autonomously in matters affecting personal well-being without explicit human confirmation.
Zero Exploitative Telemetry
When people reflect on their feelings, relationship struggles, or psychological traits, their data belongs strictly to them. It must never be weaponized for ad targeting, behavior prediction, or user profiling.
Refuse Cheap Flattery
A useful AI tool will gently highlight inconsistencies in your thoughts rather than pretending you are infallible. Constructive friction is the foundation of intellectual integrity.
Designed for Logging Off
The measure of successful software is not "daily active minutes spent scrolling." It is whether the user feels refreshed, gains clarity, and returns to their family, craft, and garden with renewed energy.