Automation and AI at the Philadelphia Inquirer

Stephen Stirling

Data Editor, Philadelphia Inquirer

KEY DISTINCTION

Automation vs. AI

Automation

Rule-based systems that follow predefined instructions to handle repetitive tasks. No learning or judgment involved.

  • If X happens, do Y
  • Consistent, predictable
  • Easy to audit and trust

AI / LLMs

Systems that learn patterns and make judgments. Powerful but require caution around accuracy and ethics.

  • Interprets and generates
  • Can surprise you (good & bad)
  • Requires human oversight

These two things are not the same. Automation can do a lot without AI.

Automation at the Inquirer

We use automation to report, analyze, and track publicly available data feeds.

APIs

Automated data pulls from public interfaces

OpenDataPhilly

City and regional open data platform

Gov Websites

Scraping and monitoring official pages

AUTOMATION EXAMPLE

Snowfall Forecast Map

How It Works

  • Pulls NDFD snowfall forecast data automatically
  • Updates every 6 hours
  • Forecast changes, but the data structure doesn't
  • Source is reliable: works 99% of the time

AUTOMATION EXAMPLE

Airport Delay Trackers

How It Works

  • Pulls data every 15 minutes from PHL website
  • Monitors arrivals and departures for delays
  • Alerts reporters automatically when disruptions spike

AUTOMATION EXAMPLE

Following ICE Locally

How It Works

  • Pulls data from Deportation Data Project
  • Triggers when new updates come in
  • Allows quick turnaround on big data dumps

AUTOMATION EXAMPLE

Sports!

How It Works

  • Pro sports has tons of structured data access
  • I use it as my experimentation playgound
  • This pulls from the NBA API and triggers when a new game ends
  • Instant analysis on every Sixers player’s game and season performance

AI at the Inquirer

The data team is helping lead experimentation with AI.

We are exceedingly cautious, but optimistic about how it can help reporters.

Don't be an asshole.

AI is a human input tool.

If you use it like an asshole, you are an asshole.

AI PROJECT

Scribe

  • There are more than 500 public meetings in the Inquirer’s coverage area each month.
  • Scribe monitors local government meetings and produces AI-generated summaries scored by newsworthiness. It helps reporters see what they can't.
  • Led by Stephen Stirling and Kevin Hoffman. Developed by a team of 18 reporters and editors.

SCRIBE

How It Works

1

Monitor

2

Transcribe

3

Score

4

Deliver

AI APPLICATIONS

Other AI Uses

Dewey

App trained on the Inquirer's archives going back decades. Easy search for prior coverage across the full library.

Coding Assistance

Several reporters use AI to get over coding hurdles. Claude Code is exceptional for this kind of technical support.

Sticky Syntax

Helpful for connecting to data sources like APIs that often have bespoke coding syntax that's hard to remember.

Lenfest Institute collaboration

GUARDRAILS

What We Don't Use AI For

Publishing Creative Content

No AI-generated words, images, or creative material is published

Data Analysis

Security concerns, and we have these skills in house

Most Things

AI is a targeted tool, not a replacement for journalism

Even if you hate AI, learn about it.

Thank You

Stephen Stirling

Data Editor, Philadelphia Inquirer