Title image above is a still taken from a video by Sabine Hossenfelder entitled “AI Slop Is Spreading In Science, Too”

First published 4th June 2025

Science has been corrupt(ed) for ages. I can’t point to a specific decade or even century, but I know I’ve been jaded since the 90s, and I came to the jaded party pretty late. The more I look into it the further back I go, and I can point to examples from the 1800s.

Quite simply, pipers (researchers) play the tunes they’re paid to play (by research grants). But it goes deeper: in a world of “publish or perish” it is only natural that an entire industry of “pay to publish” had sprung up to meet this need. Yes, whether genuine ground-breaking research or complete and utter drivel, there is always someone who will gladly publish it for you — a few thousand US dollars later, and congratulations, you are now a published researcher in a journal (degree of quality varies), who can cite your own work for more grants! Rinse. Repeat.

(I stress that the intent is not always sinister, as that link will show. This approach arose from the digital age putting financial pressure on the traditional publishing model of printing sometimes inch-thick journals posted out monthly or quarterly to subscribed academics, libraries and research institutions worldwide. But it is very much open to abuse, and it is abused.)

And now, surprise surprise, ‘AI’ (very much artificial and very much not intelligent — see below) has been adopted by lazy, corrupt, incompetent, immoral, unethical (I’m running out of adjectives) shills to ruin science even further.

This short video is well-worth watching for a quick summary of all that is going on. It’s quite shocking and the ramifications are horrifying.

And if, through no fault of your own, you do not understand how ‘AI’ really works, and you do currently believe that AI really is intelligent, I sincerely hope that the 2:50 mark on “vegetative electron microscopy” (from where the header graphic comes from) changes your mind completely. And please keep reading below for more on this.

Sabine concludes with, “But the way it currently looks, the future of science isn’t about finding truth — just about generating statistically plausible sentences about it.”

A brilliant summation, but the true significance of her words is very underplayed here.

You see, ‘AI’ language models ARE literally nothing more than the stringing together of words and phrases that statistically are highly likely to occur together. Most people do not understand this. And our “vegetative electron microscopy” friend above couldn’t be more concrete a proof of this.

Please do take the time to read this somewhat lengthy but highly informative and very easy to read article: The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con. This will give you a much better grasp on just how insidious ‘AI’ truly is.

For more on how easily we can be manipulated, please watch this video — thanks Jason! — which is 100% AI-generated. Most (not all) of the generated imagery is indistinguishable from reality without deep analysis. (It can’t generate images of words though! You can’t help but notice the appalling captions when they appear, but also look closely at any background signage that appears! All gibberish.)

I shall conclude this article with this wise and insightful comment underneath that video:

@frantz4318
You know what? As crazy as it looks, I think the world needs this... Finally we cannot trust anything online. This is maybe the only way for people to connect in person, irl. I’ll continue to enjoy hiking, biking and fishing with kids and spend less and less time online