About Splitfeed

Splitfeed is a news site for one beat: artificial intelligence. Every story we publish comes with the Split, a single screen showing who is cheering, who is alarmed, who is hedging, and who is worried about their livelihood. Same facts, four reactions, no spin from us.

Why the Split

Most AI coverage picks a side and hands you the quotes that agree with it. That leaves you with half the argument and none of the context. We think the more useful product is the disagreement itself: the strongest version of each position, attributed to the person who said it, next to the others. You decide which one holds up.

The four camps

Every reaction we collect is filed into one of four camps. The labels describe the take, not the person.

• AccelerateFaster is better. This is progress and we should lean in.
• AlarmThis is dangerous, moving too fast, or needs to be stopped.
• MiddleCautiously optimistic, nuanced, or wait-and-see.
• DisplacedFocused on jobs, livelihoods and who gets left behind.

A fifth label, Skeptic, is a flag rather than a camp. It marks takes that dismiss the story's significance altogether. Skeptic takes still sit in a camp, but they are shown with the flag and weighted separately in the Doom / Boom index.

How stories are selected

We do not write news. We read it. An automated pipeline pulls headlines from a fixed list of outlets and lab blogs using their public RSS feeds, groups duplicate coverage of the same event into one story, and scores each story by how many independent sources are covering it and how fast new coverage is arriving. A story is only published once it has reached three or more independent sources. Single-outlet stories wait.

Each published story carries a short neutral summary and a link to every article we grouped under it. The summary is drafted by a language model from the source articles and is limited to facts those sources agree on.

How quotes are selected

Reactions come from three places: public posts on X, public Reddit threads, and a curated watchlist of accounts whose takes we always pull regardless of reach. Each reaction is classified into a camp, then a human editor reviews the classification before the reaction appears on the site. We show the featured quote for each camp plus a short list of additional takes.

Every quote is an attributed excerpt with a link to the original post. We do not edit quotes beyond trimming for length, and we do not paraphrase. If a camp has no reaction to a story, we say so rather than fill the gap.

Human review

Camps are editorial calls. Software proposes them and people confirm them. If you think a take has been filed in the wrong camp, or a quote has been attributed to the wrong person, tell us and we will fix it quickly. Attribution corrections are handled ahead of everything else.

What we do not do

• We do not scrape article text. We read public feeds and link out.
• We do not invent quotes or attribute statements to people who did not make them.
• We do not sell your data. See the privacy policy.