AI Usage Policy
As a technology journalist and developer, I’m always learning new technologies and trying to stay on the cutting — if not bleeding — edge. Artificial intelligence is one of the most transformative technologies of our time, and I actively explore its capabilities as part of my work.
That said, I believe transparency matters. My basic policy is this:
I am responsible for all the content that I deliver to my clients. For my journalism, I do not use AI to create any part of the content, and I limit my use of AI services to searching the internet, helping analyze documents, and transcribing interviews. In other, non-journalism work, I may use some AI tools to help me at certain stages, such as brainstorming, but I will never directly use output from an LLM in my work. Development is a different beast, and I use several AI systems (Claude Code and JetBrain's Junie, specifically) for coding. However, I always review the code and take responsibility for the correctness of the code and its output (such as data analyses).
How I Use AI
I'm gradually adding large language models (LLMs) and AI-based services into my workflow. Here are my main use cases:
Research
- Searching the internet, but I always check facts and data.
- Summarizing long reports, but I always use those summaries to go back and read through the reports.
- Quick analyses of data to gain insight into possible trends, but I will always build my own analysis notebook to produce charts.
- Creating summaries of recent events that may be connected to what I'm covering.
- Acting as an assistant to vet ideas and analyses.
Interviews and sourcing
- Transcribing interviews and presentations, and then summarizing key points, but I always check any quotes that I use against the audio.
- Translating articles and interviews from other languages.
- Brainstorming questions for interviewees.
Code Development
- Creating new features, tests, and frameworks for my various projects.
- Helping me review code and understand other codebases.
- Helping to test possible alternative designs and refactor code.
Accuracy and Verification
AI systems can produce errors, hallucinations, or misleading outputs. Because of this:
- I verify facts using primary sources whenever possible
- I prioritize government, academic, and direct reporting sources
- I treat AI outputs as starting points, not authoritative answers
Data and Privacy
I do not knowingly put the following into any third-party AI system:
- Confidential information
- Off-the-record material
- Sensitive source data
I use local LLMs when directly interacting with sensitive or confidential data.
Evolving Approach
AI is evolving rapidly, and so is my use of it. This policy may be updated as tools, standards, and best practices develop.