SciBot

SciHub is the world's largest collection of pirate academic papers, liberated from the parasitic academic journals.

It has now launched a chatbot which puts 15 years of its collection of good academic research behind a natural language front-end, and presumably some kind of RAG.

https://sci-bot.ru/

This is something, an AI that is fed only on real scientific knowledge, that the world's epistemic ecosystem obviously needs. I've been begging everyone I know in academia, for the last three years, to realise this and to try to push for something to be done about it.

We are in a crazy world where real academic knowledge is hidden behind paywalls, while everyone uses chatbots full of nonsense scraped from social media. People complain about lazy students. They complain about AI. They preach abstinence. But the one thing no-one seems to want to do to fix the problem is to make an AI that specialises in good academic knowledge.

Universities don't / won't / can't do this. The academic journals certainly won't. Google aren't doing it. Nor are the other AI companies who are nevertheless happy to pretend that AI is a boon to education.

In fact, AFAICT, you need to be an outlaw, hiding away in central Asia, buffered from the Western liberal world order and its copyright regime, by its three most hated enemies - Russia, China and Iran - to be able to make the one academic / AI tool that the world really needs.

I like academia. I don't want to say it's failed. Academics have their own problems and their own struggles. But looking at it, during this current AI revolution, I have to say that it's failing. Just as it failed to adapt to the web.

Academia is failing because it can't figure out how to adapt to the free-flow of information that the internet unleashed, and now the rise of AI.

The key to that adaptation is to recognise that its high-quality knowledge needs to be as freely and easily accessible to people as the rival low-quality knowledge and nonsense which circulates everywhere.

That involves tearing down the paywalls (and, really, cutting the connection to the journals who benefit from those paywalls). It involves changing the formats of academic writing to be more accessible and convenient for a wider audience. And it involves embracing chatbots / LLMs and the natural language UX that everyone is going to want for computers and knowledge from here on out.

We are on the verge of a new dark age, as the quality knowledge academia generates is swamped by nonsense. If anything of our current epistemic ecosystem survives, it will be largely thanks to AlexandraElbakyan and people like her, who are actually saving our academic culture from itself.