John Furcean

Google has an internal "let it break" essay about a hero engineer whose hard work ends up being a net negative (by masking the underlying issues). My manager sent me that essay when I was trying too hard to get the collective TensorFlow unit test suite green.

.@msunatsci students Lauren Jin and Aswath Karai, have been awarded the Barry Goldwater Scholarship, one of the nation’s top honors for undergraduate researchers pursuing careers in STEM. 💚🤍 Read more about their journey and research here: studentawards.msu.edu/news/two-msu-s…

Seven @michiganstateu College of Natural Science Faculty have been recognized as Fellows of the American Association for the Advancement of Science (@aaas). Read about these talented researchers: msutoday.msu.edu/news/2026/03/9…

New MSU microrobots are designed to be guided through the body, tracked in real time, and activated to target treatment where it is needed with the potential to reduce patient risks and speed healing. bit.ly/4rPXtbz

Glaucoma is the silent thief of sight because it damages the optic nerve. Wen Li, professor in the ECE, is part of a team that pioneered a new contact lens-based monitoring system promises to improve the diagnosis & treatment. bit.ly/4rWDN5Y

SOC 2 is about to get tested and that's not just because of Delve. It's AI. I can't even begin to tell you how many teams want to use tools that are no where near compliant and navigating that is really difficult because no one wants to say no to innovation or productivity of course. No one knows what the tools do either. So you end up in these situations where there's a lot of explaining required and a lot of risk assessment required all while the AI landscape and tools are moving really quickly. It's a nightmare to be frank. I fully expect companies to end up with massive, massive, security issues because it's also incredibly difficult to police. Take this one for example. Claude Cowork. In tech preview. Not covered by the same compliance/DPA/etc. as their other products. You can't control access to it at a granular level. It's account-wide on/off AND it's on by default (if I'm not mistaken). So as these services gain new features (that apparently can be conveniently out of scope, but somehow enabled by default), you end up in a trap. Organizations have to constantly review the sand shifting under their feet. Obviously we've seen issues with OpenClaw and LiteLLM, etc. This is all just the tip of the iceberg I'm afraid...Don't get me wrong, I love AI, it's a powerhouse for productivity IF (the biggest word in the world here) used properly and responsibly. That's a pretty tall order.

Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: goo.gle/4bsq2qI

Two distinguished Michigan State University professors, Kevin C. Elliott and John C. Besley, are part of a new multi-institutional NIH-funded research hub at Texas A&M University, focused on transforming how industrial and consumer-use chemicals are evaluated for human safety.

A new study from @MSUNatSci's Shannon Manning could explain why some mothers can still pass Group B Streptococcus, or GBS, to their babies after childbirth even when they’re treated with antibiotics.

For 20+ years, @Google has been dedicated to OSS and the developers who secure it. To combat the current threat landscape, we’re joining our industry peers to make new commitments. blog.google/innovation-and…

Researchers spend hours and hours hand-crafting the strategies behind LLM-driven optimization systems like AlphaEvolve: deciding which ideas to reuse, when to explore vs exploit, and what mutations to try. 🤖But what if AI could evolve its own evolution process? We introduce EvoX, a meta-evolution pipeline that lets AI evolve the strategy guiding the optimization. It achieves high-quality solutions for <$5, while existing open systems and even Claude Code often cost 3-5× more on some tasks. Across ~200 optimization problems, EvoX delivers the strongest overall results: often outperforming AlphaEvolve, OpenEvolve, GEPA, and ShinkaEvolve on math and systems tasks, exceeding human SOTA, and improving median performance by up to 61% on 172 competitive programming problems. 👇

Announcing Copilot Cowork, a new way to complete tasks and get work done in M365. When you hand off a task to Cowork, it turns your request into a plan and executes it across your apps and files, grounded in your work data and operating within M365’s security and governance boundaries.

Come out for an evening of fun as we support Give Green Day on 03.10.26! Your Mid-Michigan Spartans are championing three giving opportunities that support MSU! Please click on the image below for more details & event information! ⬇️ #GoGreen eventbrite.com/e/mid-michigan…

Prof. Donald Knuth opened his new paper with "Shock! Shock!" Claude Opus 4.6 had just solved an open problem he'd been working on for weeks — a graph decomposition conjecture from The Art of Computer Programming. He named the paper "Claude's Cycles." 31 explorations. ~1 hour. Knuth read the output, wrote the formal proof, and closed with: "It seems I'll have to revise my opinions about generative AI one of these days." The man who wrote the bible of computer science just said that. In a paper named after an AI. Paper: cs.stanford.edu/~knuth/papers/…

Dive into powerful storytelling with The #Waterpeople Podcast 🌊 Hosted by surfer & writer Lauren L. Hill and wave rider Dave Rastovich who chat with some of the most engaging waterpeople on the planet 🌏 Listen 🎧 waterpeoplepodcast.com/episodes/ #MSUWaterAlliance #OceanConservation

From creating a race car safety device that protects drivers from injury to revolutionizing chemotherapy, Spartans have contributed to more than 3,300 inventions. #SpartansWill

This week's MGI Research Seminar was presented by Asst. Prof. @MeghanMilbrath of MSU's dept. of entomology. She spoke about bacterial brood diseases in honey bees. Thank you, Dr. Milbrath!

In late January 2025, Mywish Maredia was traveling in Myanmar for research when she received a troubling email. The federal government had cancelled Maredia and her colleagues’ award, and was ordering them to stop working. The next day she was on a plane back to Michigan, leaving her work unfinished. statenews.com/article/2026/0…

Bytedance just dropped a paper that might change how AI thinks. Literally. They figured out why LLMs fail at long reasoning — and framed it as chemistry. The discovery: Chain-of-thought isn't just words. It's molecular structure. Three bond types: • Deep reasoning = covalent bonds (strong, unbreakable) • Self-reflection = hydrogen bonds (flexible, context-aware) • Exploration = van der Waals (weak, ever-present) Why most AI "thinking" sucks: Everyone's been imitating keywords — "wait," "let me check" — without building the actual bonds. It's like copying the shape of a protein without the atomic forces holding it together. Bytedance proved: structure emerges from training, not prompting. The fix: Mole-Syn Their method doesn't just generate text. It synthesizes stable thought molecules. Results: better reasoning, more stable RL training. Bytedance is treating AI reasoning like organic chemistry — and it works. Paper: arxiv.org/abs/2601.06002

24 dedicated people. $30M spent on development. Extreme specialization, speed, and power efficiency. Today we launch Taalas’ first product. Check it out: Details: taalas.com/the-path-to-ub… Demo chatbot: chatjimmy.ai API: taalas.com/api-request-fo…