The American Conservative’s AI Writing Policy
TAC is committed to preserving the trust it has built up with its readership over a quarter-century.
IBM had a famous dictum in its 1979 training manual: “A computer can never be held accountable, therefore a computer must never make a management decision.”
As far as corporate training-manual maxims go, we think that’s a pretty good one, and generalizable to our own work. Journalism is useful insofar as it is trustworthy, and it is trustworthy only insofar as a person or set of persons are responsible for getting facts and analysis right. Failure must come at the risk of public shame and, in extremity, professional jeopardy.
The only way a journalist can be held responsible for the entirety of an article is for him to write the entire thing, including the background material that nonprofessionals may think is mere window-dressing for the central reporting or argument. Any reporter can regale you with war-stories in which perfunctory fact-finding for context or kickers brought new detail or insight to light, radically changing the entire premises, emphases, or framing of the article.
Journalistic writing is the artifact of the writer’s critical engagement with fact; if you are not engaging with the facts critically as you put words onto the page, you are not doing journalism. Putting your name on an article is an assertion that you have actually thought through what you have written, and will stand by it.
For this reason, The American Conservative does not accept stories containing AI-generated prose. While AI has opened exciting vistas of technical progress, including for the research and data-processing that go into journalism, it is not an appropriate tool for the replacement of human judgment, which is what bylined writing represents. A computer can never be held accountable. No matter how proficient AI tools become at generating aesthetically acceptable prose, they cannot supply authority and responsibility.
TAC’s editors and writers are not infallible, but we’d like to think that over nearly a quarter-century they’ve amassed a pretty good track record, and have earned a reputation for independence and honesty. We do not present you, our readers, with articles we have not thought through ourselves and we are not willing to stand behind. To maintain this standard of responsibility, our editors check every submission not only using their own judgment but also the best available tools for AI detection. We will not publish those we believe to be generated by AI.
We believe the American public is with us on this issue. Ninety-four percent of those surveyed by Pew Research think it is important for them to know whether a given piece of media is generated by AI. Just 10 percent believe that AI use will have a positive effect on journalism. Fifty-six percent say they are less confident in an article composed using AI. These are not ambiguous numbers. Americans understand that journalism is ultimately a matter of trust, and trust is a matter of human responsibility.
We here at TAC value your trust more than anything; it is the fundamental currency of our work. If you think our project is worthwhile, please consider making a tax-deductible donation. As always, we are deeply grateful for your support and your loyalty, which are remarkable.
Curt Mills
Executive Director
Jude Russo
Managing Editor
The post The American Conservative’s AI Writing Policy appeared first on The American Conservative.
