These three controversies from Wikipedia’s past reveal how genuine conversations can achieve—after disagreements and controversy—compromise and evolution of Wikipedia’s features and formats. Reflexive vetoes of new experiments, as the Simple Summaries spat highlighted last summer, is not genuine conversation.

Supplementing Wikipedia’s Encyclopedia Britannica–style format with a small component that contains AI summaries is not a simple problem with a cut-and-dried answer, though neither were VisualEditor or Media Viewer.

Surely, AI summaries are exactly the same as stuff like VisualEditor and Media Viewer, which were tools that helped contributors improve articles. Please ignore my rhetorical sleight of hand. They're exactly the same! Okay, I did mention AI hallucinations in one sentence, but let's move on from that real quick.

A still deeper crisis haunts the online encyclopedia: the sustainability of unpaid labor. Wikipedia was built by volunteers who found meaning in collective knowledge creation. That model worked brilliantly when a generation of internet enthusiasts had time, energy, and idealism to spare. But the volunteer base is aging. A 2010 study found the average Wikipedia contributor was in their mid-twenties; today, many of those same editors are now in their forties or fifties.

Yeah, because Wikipedia editors are permanently static. Back in 2001, Jimmy Wales handpicked a bunch of teenagers to have the sacred title of Wikipedia Editor, and they are the only ones who will ever be allowed to edit Wikipedia. Oh wait, it doesn't work like that. Older people retire and move on, and new people join all the time.

Meanwhile, the tech industry has discovered how to extract billions in value from their work. AI companies train their large language models on Wikipedia’s corpus. The Wikimedia Foundation recently noted it remains one of the highest-quality datasets in the world for AI development. Research confirms that when developers try to omit Wikipedia from training data, their models produce answers that are less accurate, less diverse, and less verifiable.

Now that we have all these golden eggs, who needs the goose anymore? Actually, it is Inevitable that the goose must be killed. It is progress. It is the advancement of technology. We just have to accept it.

The irony is stark. AI systems deliver answers derived from Wikipedia without sending users back to the source. Google’s AI Overviews, ChatGPT, and countless other tools have learned from Wikipedia’s volunteer-created content—then present that knowledge in ways that break the virtuous cycle Wikipedia depends on. Fewer readers visit the encyclopedia directly. Fewer visitors become editors. Fewer users donate. The pipeline that sustained Wikipedia for a quarter century is breaking down.

So AI is a parasite that takes from Wikipedia, contributes nothing in return, and in fact actively chokes it out? And you think the solution is for Wikipedia to just surrender and implement AI features? Do you keep forgetting what point you're trying to make?

Meanwhile, AI systems should credit Wikipedia when drawing on its content, maintaining the transparency that builds public trust. Companies profiting from Wikipedia’s corpus should pay for access through legitimate channels like Wikimedia Enterprise, rather than scraping servers or relying on data dumps that strain infrastructure without contributing to maintenance.

Yeah, what a wonderful suggestion. The AI companies just never realized all this time that they could use legitimate channels and give back to the sources they use. It's not like they are choosing to do this because they have no ethics and want the number to go up no matter the costs to themselves or to others.

Wikipedia has survived edit wars, vandalism campaigns, and countless predictions of its demise. It has patiently outlived the skeptics who dismissed it as unreliable. It has proven that strangers can collaborate to build something remarkable.

Wikipedia has survived countless predictions of its demise, but I'm sure this prediction of its demise is going to pan out. After all, AI is more important than electricity, probably.