Sci-Hub Launches Sci-Bot: An AI Assistant Built on 88 Million Open Research Papers

Forum topic · QianXun · 2026-05-16

Summary

Sci-Hub, the controversial free paper-access platform created by Alexandra Elbakyan in 2011, has announced a new AI tool called Sci-Bot. While Sci-Hub bypasses academic paywalls to provide free access to over 88 million scientific papers, Sci-Bot acts as a retrieval-based assistant over that corpus: it answers scientific questions only from the text of stored papers, refuses to fabricate claims, and attaches direct links to the source PDF for verification. The post frames this as an evolution from simply unlocking papers to making them searchable and understandable, contrasting Sci-Bot's citation-grounded answers with the hallucination problem of general-purpose chatbots like ChatGPT. It also recaps Sci-Hub's dual reputation: facing tens of millions of dollars in lawsuits from publishers while serving as an essential resource for researchers, including those at well-funded institutions lacking access to specific databases. Sci-Bot is available at https://sci-bot.ru/. The author argues that combining open access with grounded AI search could accelerate scientific progress by removing both financial and cognitive barriers to knowledge.

Why Sci-Hub Built an AI

The original post (in Chinese) tells the story of Sci-Hub and its newly launched AI companion, Sci-Bot.

The Problem: Knowledge Behind Paywalls

Humanity's collective research—cancer treatments, cosmology, new energy—is locked behind paywalls, where a single 10-page paper can cost around $35. For a doctor in a poor region or an unfunded independent researcher, that price is an insurmountable barrier.

In 2011, Kazakhstan-born developer Alexandra Elbakyan created Sci-Hub, often described as the "Robin Hood of science," to bypass these barriers.

Sci-Hub: More Than a "Pirate Site"

The post frames Sci-Hub as eliminating informational entropy: it has liberated over 88 million scientific papers and made them freely available worldwide. Its reputation is deeply contradictory:

Legally: a notorious "pirate," facing lawsuits worth tens of millions of dollars.
Practically: an indispensable lifeline—even professors at Harvard or Peking University occasionally rely on it when their institutions lack a specific database.

2026: The Birth of Sci-Bot

Liberating papers was only step one; nobody can read 88 million of them. So in March 2026, Elbakyan gave the "master key" a new brain: Sci-Bot ( https://sci-bot.ru/ ).

If Sci-Hub is an infinite library, Sci-Bot is an omniscient librarian. Unlike general-purpose AI (e.g., ChatGPT), which often hallucinates because it can't cite specific sources, Sci-Bot works differently:

1. Read first, answer second: it searches across all 88 million papers in Sci-Hub when answering a question. 2. No fabrication: every answer must be grounded in the original paper text; if a paper doesn't say it, the bot won't invent it. 3. Instant verification: every claim in an answer links directly to the source PDF.

Why It Matters

The author invokes Feynman's principle—"I have to think from scratch to understand what I'm doing"—arguing that Sci-Hub and Sci-Bot protect humanity's right to think from first principles. When knowledge is no longer a luxury, science may shift from linear to exponential progress.

The takeaway: Sci-Hub takes the books out of the vault; Sci-Bot teaches you to read them. No intelligence on this planet should be buried because of poverty.