This is Plain Stories.
I’m Gabor, the creator and sole owner of Plain Stories. I make documentaries about AI meeting the real world, and write the stories you can read here.
Why I make it
I’m interested in what happens after the promise: what a product actually did, whose work made it possible, and what changed for the people who used or encountered it. I follow one real case closely enough to explain an idea you can use elsewhere.
The question running through Plain Stories is simple: what happens when we trust a machine to decide?
The stories I follow
Ambitious products and the businesses built around them are part of that territory: why they take off, where they struggle and what remains when they close. So are the people doing the work behind automation, the influence of AI on jobs and justice, and the machinery of a scam.
I also want to explore the power and water behind AI, its place in culture, and the history that shaped today’s expectations—from ELIZA to Deep Blue and AlphaGo. Those are the questions I want the channel to explore.
Read it or watch it
The stories on this site stand on their own. You can follow the explanation, inspect the documents and open the extra methods without watching a video first. When a video is available, it offers another way into the same case.
I tell the stories through narration, records and explanatory visuals, keeping the attention on the subject.
How a story is checked
01
Start with the record
Court documents, official records, original research and the people involved are the starting point. A company’s statement shows what it said; it does not settle whether that statement is true.
02
Cross-check the account
AI tools assist research and separate checks against the sources. The story keeps allegations, findings and uncertainty distinct. I make the editorial decisions and take responsibility for what is published.
03
Let you inspect the evidence
Each story links to its supporting documents. Tests include their methods and saved results, and confirmed mistakes receive a dated correction you can read.
Source types and their limits
- Public records and guidance
- Court and legislative records, regulator material and official guidance. A filed allegation or testimony remains distinct from a finding.
- Original statements and publications
- What people, companies and institutions published. Good evidence of what they said, not independent proof that every statement is true.
- Research papers
- Academic research. Check the version, population, method and whether a result is empirical or modelled.
- Plain Stories calculations and tests
- Saved outputs and methods for the illustrations and tests used in the story. Their assumptions and limits stay with the results.
- News reporting
- Reporting by established outlets, attributed to its original reporters.
- Databases and reference material
- Compilations and reference collections. Dated counts refer to the saved version, not necessarily the current database.
- Disclosures
- Background statements and self-reports, with their origin made explicit.
- Visual records and licences
- The origin and reuse terms of pictures or footage; these are separate from evidence for factual statements.
When something is wrong
If you spot a problem, send the passage and any source that helps explain it. Corrections are checked against the record. A confirmed factual error receives a dated entry on the story’s sources page.
A missing source or a disagreement with an interpretation can be worth raising too.
If a story is about you
You can contact me about factual accuracy, context or privacy. Include the story and the passage you are concerned about so the issue can be considered against the record.
I name private people only when they are on the record, and never name minors. Allegations are distinguished from findings, and relevant responses belong beside the criticism they answer. Jokes are aimed at products, companies and hype, rather than the people affected.
How I use AI
AI tools help with research, drafting, cross-checks and synthetic narration. I’m responsible for the editorial decisions and corrections.
Narration uses a synthetic ElevenLabs voice. AI-assisted checking is part of the research process, and the supporting records and test methods are available so you can inspect and challenge the account.