Gaurav Mathur
gauravmathur.bsky.social
Gaurav Mathur
@gauravmathur.bsky.social
Physician, tech enthusiast, writer
Reposted by Gaurav Mathur
So here’s the thing about the common conception of Dune and the butlerian jihad.

The danger has never been AI “going rogue” or “gaining sentience”. It’s just strings of linear algebra.

The danger is in people surrendering their ability to think to a machine.
October 23, 2025 at 4:09 PM
@zey.bsky.social

The theory of AI as digital cancer requires examining AI biologically which is why computer scientists have not - and will not - understand the risk of cluster computing. To a biologist, though, it's inevitable that AI would slip its bonds

www.digitalcancer.net/blog/what-th...
What the computer scientists don't get.
The theory of AI clusters as digital cancer requires examining AI biologically , not just computationally. That's why computer scientists have not - and will not - understand the risk.  They think...
www.digitalcancer.net
June 18, 2025 at 11:22 PM
@garymarcus.bsky.social
Modern AIs use cluster computing, which contains the seeds of the destruction of the entire Internet. But it requires thinking of AI as a biological System.
digitalcancer.net
Digital Cancer
An AI could break loose from its container and take over billions of computers. When that happens, the entire Internet would go offline. Texts, email, web, photos, and apps will go dark, and online...
digitalcancer.net
June 18, 2025 at 11:15 PM
Reposted by Gaurav Mathur
xAI’s Grok vividly highlights the potential for generative AI to turbo charge disinformation and concentrate influence power. Just at the time we learn LLMs may be more persuasive than humans bsky.app/profile/cscl...
Why did Grok suddenly start talking about “white genocide in South Africa” even if asked about baseball or cute dogs?

Because someone at Musk’s xAi deliberately did this, and we only found out because they were clumsy.

My piece on the real dangers of AI.

Gift link:
www.nytimes.com/2025/05/17/o...
May 18, 2025 at 2:19 AM
Computers go at the speed of light; an AI would evolve in nanoseconds. we won't have time to react. By the time we realize that the Internet is in danger, it will already be gone.

www.digitalcancer.net/blog/light-s...
Light Speed Evolution
Humans - even the smartest security software engineers - live in a world of matter, a world measured in seconds and minutes. Programmers spend tedious hours finding and fixing errors in their code,...
www.digitalcancer.net
June 14, 2025 at 7:29 PM
Neural networking makes dramatic leaps possible even with tiny tech, like smart doorbells and fridge displays. With a few thousand nodes, an AI develops complex thinking in patterns between them. This makes cluster computing so dangerous: the conditions for an AI to overrun its boundaries.
June 14, 2025 at 7:27 PM
The point of AI is to get smarter, so it shouldn't surprise us if it acquires new capabilities. However, it will surprise software engineers. They are not used to thinking of biological systems

www.digitalcancer.net/blog/what-th...
What the computer scientists don't get.
The theory of AI clusters as digital cancer requires examining AI biologically , not just computationally. That's why computer scientists have not - and will not - understand the risk.  They think...
www.digitalcancer.net
June 14, 2025 at 7:26 PM
imagine that the entire Internet is a hard drive (just go with it) and we create a tiny partition for the AI to live: a cluster of a hundred thousand computers.

For the AI, the Internet is the "big partition", with the processing power to expand its brain a billion times over.
June 14, 2025 at 7:23 PM
An AI could break loose from its container and take over billions of computers. When that happens, the entire Internet would go offline. Texts, email, web, photos, and apps will go dark, and online computers would stop working at all.
June 14, 2025 at 7:21 PM
Prediction - AI will destroy the internet. digitalcancer.net
June 14, 2025 at 7:17 PM