terça-feira, 3 de junho de 2025

Why a Former MIT Professor Released 3,800 Papers Over The Past Two Years ?

Douglas C. Youvan, now 70, earned a PhD in biophysics from the University of California, Berkeley in 1981, where he conducted pioneering research in photosynthesis. He later served as an associate professor of chemistry at MIT, specializing in the spectral analysis of photosynthetic bacteria. Over the course of his scientific career, he authored 66 peer-reviewed articles indexed in the Scopus database, the last one was published in 2001.

But then, things took an unusual turn. Starting in March 2023 and continuing through May 18, 2025, Douglas 
Youvan embarked on an obsessive ResearchGate posting spree. He uploaded multiple entries each day—covering quantum physics, AI, mathematics, philosophy, theology, and ethics—and ultimately amassed an astonishing 3,800 publications.

What makes the story more intriguing is a recent essay in Big Think, where five respected senior scientists — Brian Swimme, Marcelo Gleiser, Alan Lightman, Nahum Arav, and Joel Primack — reflect on how modern cosmology shapes our understanding of meaning, ethics, and human identity. 

In a piece where the word spiritual appears seven times, the tone is unmistakably contemplative. Physicist Alan Lightman—formerly at Harvard and now a professor of the practice of the humanities at MIT—offers an insight worth quoting: "We...are the only way the Universe can observe itself ”

Let´s not forget that several renowned physicists - Heisenberg, Pauli, Schrödinger and even Einstein - spent their later years striving to bridge rigorous scientific work with life’s existential questions that lie beyond empirical proof.

Curiously, a concept proposed in one of Youvan’s recent publications — The Æthervault Nexus — appears to echo many of the same themes, suggesting that meaning emerges through complex phenomena that unite spirituality, cosmology and quantum mechanics.

segunda-feira, 2 de junho de 2025

A mentira é a nova verdade: Assim escreve um catedrático convidado da Nova SBE



Para lá dos dois artigos mencionados no post supra, o semanário Expresso continha também um artigo bastante interessante, com autoria de um catedrático convidado da Nova SBE, do qual reproduzo o seguinte excerto, que nesse artigo se segue a um alerta de uma investigadora da Universidade de Stanford: 

No presente contexto entendo como especialmente pertinente recordar um post anterior de Janeiro de 2024, sobre um singular artigo publicado na conhecida revista The Economist, que alterou para a dificuldade em impedir a falsificação da realidade, num mundo dominado pela IA, o que é suficiente grave para levar a uma alteração significativa da natureza da missão universitária https://19-pacheco-torgal-19.blogspot.com/2024/01/the-economistai-generated-content-is.html

PS - Hoje mesmo no jornal Público, o antigo Presidente do IST e actual Presidente do INESC, num artigo particularmente oportuno, questiona se pretendemos uma abordagem da IA, que valoriza a supervisão humana, o juízo crítico e a responsabilidade ética (centauro) ou em alternativa uma fusão profunda entre o ser humano e a IA, onde a cognição se torna híbrida, com interfaces cérebro-computador e simulação de mentes humanas em computador (ciborgue). A resposta a essa pergunta é mais ou menos evidente, pois se numa democracia a primeira opção é a única que faz sentido, já nos países onde vigoram sistemas ditatoriais é muito provável que optem pela segunda. 

domingo, 1 de junho de 2025

AI’s Bifurcated Brilliance: Physics Phenom, Civil Engineering Underachiever



Recent research from Germany demonstrates that GPT-4o and o1-preview possess remarkable problem-solving capabilities on Olympiad-level physics tasks (paper linked above), frequently outperforming human competitors. 

In contrast, a study by researchers from New Zealand and the US examined ChatGPT-4o and a Custom GPT on first-year civil engineering statics problems. Unlike their performance on theoretical physics challenges, the AI models struggled with nuanced engineering questions—such as distinguishing tension from compression and resolving inclined forces—and exhibited limited proficiency in interpreting diagrams. These shortcomings highlight that real-world engineering relies on iterative, collaborative workflows—peer review, simulations, and safety factors—rather than isolated, purely theoretical problem-solving.  https://arxiv.org/abs/2502.00562

PS - A more impactful application of ChatGPT in civil engineering lies in assisting with compliance with the strict codes and standards that govern design projects, ensuring they meet required levels of safety, reliability, and efficiency. Checking that every design follows all the detailed regulations can be really complex and time-consuming—it’s usually something expert engineers do manuallyA recent paper introduces an open-source framework that uses large language models (LLMs) to provide engineers with accurate, code-related answers from natural language queries, including references to relevant code sections. The initial implementation, based on the National Building Code of Canada, shows promise demonstrating the potential of this approach to streamline design verification tasks. https://ascelibrary.org/doi/10.1061/JCCEE5.CPENG-6037#con1