0:00 Thirty-two percent of new scientific papers on arXiv are now flagged as being written by AI.0:06 Last week we touched on AI breakthroughs, but this...0:09 this is a different kind of metric entirely, and it's the story everyone on Hacker News is wrestling with today.0:17 What happens when the firehose of scientific knowledge gets hooked up to a content-generation machine?0:24 We're going to dig into that.0:26 But first, let’s sweep the other major threads.0:29 Because while a third of science papers might be getting automated, the tools for creation and control are also getting a LOT better.0:38 First, on the AI creation front, Qwen-Image-3.0 just dropped.0:42 This is a new image generation model from Alibaba's cloud division, and the key word here is "useful." They're not just chasing photorealism.0:52 They're chasing complex layouts.0:54 It can take a prompt that's over four thousand tokens long—that's like a short essay—and generate a complex three-by-three infographic in a single pass.1:04 The examples show it handling requests for scientific diagrams and cultural comparisons side-by-side in the same image.1:12 It’s a step towards image generation as a real productivity tool, not just a toy.1:18 Then, on the complete opposite end of the spectrum, an AI just disproved a famous mathematical conjecture.1:25 The Erdős Unit Distance conjecture, a problem in discrete geometry that's been around for decades.1:31 A proof was generated by a large language model and then—here's the key part—it was autoformalized into a machine-checkable language called Lean.1:41 Fields Medallist Mike Freedman called it "breakthrough LLM-generated mathematics being formalized in real time." So, AI isn't just writing papers, it's also...1:52 you know, solving actual math problems.1:55 Meanwhile, the tools to actually use all this stuff are getting seriously powerful, and personal.2:01 A new open-source app for macOS called Nativ scored big on Hacker News.2:06 It lets you run frontier AI models—from Google, Cohere, Liquid AI—locally on your Apple Silicon Mac.2:13 No subscriptions, no cloud dependency, no sending your data off to some server.2:18 The developer's manifesto is all about transparency: "Every line.2:22 Every model loader...2:24 No VC roadmap.2:25 No enterprise tier." It's a purist's dream for running AI on your own terms.2:30 And for those who want AI to just...2:32 do their work for them, there's Kimi Work.2:35 This is a new AI desktop agent that's designed to automate knowledge work.2:40 It's not just a chatbot.2:42 It can schedule tasks to run 24/7, browse the web autonomously to complete multi-step research, and even coordinate a swarm of specialized agents to generate reports.2:53 It has native integration with global stock market data.2:57 This is the vision of the AI assistant that works quietly in the background, ticking off your to-do list while you sleep.3:05 And in a story that has nothing to do with AI and everything to do with ownership, the open-source media server Jellyfin is having a moment.3:15 The context here is that its main commercial competitor, Plex, just jacked up the price of its Lifetime Pass to seven hundred and fifty dollars.3:24 As one commenter put it, that goes against the whole purpose of hosting your own media library.3:31 So people are flocking to Jellyfin, the free, open-source alternative.3:35 The timing is a little dramatic, as Jellyfin's original founder just announced he's stepping down from the project.3:43 But the community seems energized to carry it forward, even with the usual open-source complaints about bugs and setup complexity.3:52 Finally, for the web developers, a gorgeous new UI library called Jelly UI is making the rounds.3:58 It's a dependency-free set of web components that brings soft-body physics to native HTML form controls.4:06 Think buttons and sliders that feel tactile and squishy.4:09 It's accessible, supports dark mode, and looks incredibly fun to use.4:14 It’s just a nice reminder that even as we debate the grand future of AI, people are still out there making the web a more delightful, tangible place.4:24 Okay, let's go back.4:25 I want to pull on two of those threads, because they seem to point in completely opposite, and equally important, directions.4:34 The flood of thirty-two percent AI-generated papers on arXiv, and the single AI-generated proof for a major math problem.4:42 Let's start with the flood.4:44 A new analysis on the Unslop dot run blog looked at over twelve thousand papers on arXiv, the main pre-print server for physics, math, and computer science.4:54 They used an AI detector calibrated to have a very low false-positive rate—about zero point four percent.5:02 And what they found is staggering.5:04 Through 2021 and 2022, the number of flagged papers was flat, right at that false-positive baseline.5:11 Then, within months of ChatGPT's release, it lifts off.5:14 And now, for the most recent quarter, it's at thirty-two percent.5:19 For computer science papers, it's sixty-five percent.5:22 Two-thirds.5:23 So what does this mean?5:25 The immediate reaction is, "Oh my god, science is being replaced by bots." But it's more complicated than that.5:32 A lot of this is probably not fraudulent.5:35 It's researchers using AI as a writing assistant—to polish their English, to structure their abstracts, to write boilerplate code.5:44 It's a productivity boost.5:45 But here's the problem.5:47 Where have we seen this pattern before?5:50 To me, this looks exactly like the early 2010s and the rise of SEO content farms.5:55 You remember this.5:56 Companies realized they could use cheap labor, and later, simple algorithms, to churn out thousands of low-quality articles optimized for Google search terms.6:07 The goal wasn't to inform.6:09 The goal was to rank.6:10 To capture traffic.6:11 And for a while, it completely broke search.6:14 You'd look for a product review and get ten pages of meaningless, keyword-stuffed garbage.6:21 The arXiv situation feels like the academic version of that.6:25 It's a system being flooded with low-cost, plausible-sounding text.6:29 The incentive is to publish.6:31 More papers, more citations, better career.6:34 AI dramatically lowers the cost of producing a paper.6:38 So, people produce more papers.6:40 And here's where the analogy with SEO breaks down, and why this is so much more serious.6:46 The stakes for scientific literature are infinitely higher than for a blog post about the best blenders.6:53 Science is a cumulative enterprise.6:55 Each discovery builds on the last.6:57 If the foundation becomes polluted with plausible-sounding but subtly incorrect, or unoriginal, or just plain vacuous text, the whole structure is at risk.7:08 How do you trust a citation when the paper it points to might have been written by a machine that hallucinates facts?7:16 The signal-to-noise ratio is degrading in real time.7:19 It creates a massive verification burden that the system just isn't designed to handle.7:25 But then.7:26 THEN you have the other story.7:28 An AI that helps disprove the Erdős Unit Distance conjecture.7:32 This isn't just an AI writing a summary of existing research.7:36 This is different.7:37 According to the team at Logical Intelligence who did the work, the process started with an LLM generating the idea for the proof.7:46 It found a novel construction.7:48 That's the creative leap.7:50 But an idea isn't a proof.7:51 A mathematical proof has to be rigorous, logical, and flawless.7:56 So the next step was formalization.7:58 They used AI tools to translate that human-readable, English-language proof idea into Lean, which is a formal proof assistant.8:07 Think of it like a compiler for logic.8:09 It checks every single step and guarantees that there are no gaps, no mistakes.8:15 If the Lean code compiles, the proof is correct.8:18 And it did.8:19 This is the pattern of augmentation, not replacement.8:22 We've seen this before, too.8:24 Think about the invention of the pocket calculator.8:27 It didn't make mathematicians obsolete.8:30 It freed them from the drudgery of arithmetic to think about higher-level problems.8:36 Then came computer algebra systems like Mathematica, which could handle complex symbolic manipulation that would take a human a lifetime.8:45 They opened up new fields of inquiry.8:47 This feels like the next step in that same progression.8:51 But with a critical twist.8:53 The calculator just did the arithmetic you told it to.8:57 Mathematica just did the algebra you told it to.9:00 This new generation of AI tools is starting to suggest what to prove.9:05 It's providing the creative spark, not just the mechanical verification.9:09 It's a partner in discovery.9:11 So what does it all add up to?9:13 On one hand, you have AI as a firehose, burying us in synthetic text and threatening the integrity of our shared knowledge base.9:22 On the other hand, you have AI as a precision instrument, a microscope for logic that's helping us make genuine, world-class discoveries that were out of reach before.9:33 It's not a contradiction.9:35 It's two sides of the same coin.9:37 The technology is just a powerful lever.9:40 It amplifies whatever you point it at.9:42 Point it at the "publish or perish" incentive system, and you get a flood of mediocre papers.9:49 Point it at a grand challenge in mathematics, with a team dedicated to rigorous verification, and you get a breakthrough.9:57 This is the core tension of the week, and probably of the next decade.10:02 We have built tools that are incredibly good at generating things that look like human output.10:08 Images, papers, proofs.10:10 And that sets up a new reality for anyone doing knowledge work.10:14 The bottleneck is shifting.10:16 It's no longer about who can generate content the fastest.10:20 It's about who can verify it.10:21 The most valuable skill is no longer creating the answer, but knowing how to validate it.
