Lifeboat Foundation

Large language models (LLMs) are impressive technological creations but they cannot replace all scientific theories of cognition. A science of cognition must focus on humans as embodied, social animals who are embedded in material, cultural and technological contexts.

There is the technological question of whether computers can be intelligent, and also the scientific question of how it is that humans and other animals are intelligent. Answering either question requires an agreement about what the word ‘intelligence’ means. Here, I will both follow common usage and avoid making it a matter of definition that only adult humans could possibly be intelligent by assuming that to be intelligent is to have the ability to solve complex and cognitively demanding problems. If we understand intelligence this way, the question of whether computers can be intelligent has already been answered. With apologies to Dreyfus and Lanier, it has been clear for years that the answer is an emphatic ‘yes’. The recent advances made by ChatGPT and other large language models (LLMs) are the cherry on top of decades of technological innovation.

Researchers confront a formidable challenge within the expansive domain of materials science—efficiently distilling essential insights from densely packed scientific texts. This intricate dance involves navigating complex content and generating coherent question-answer pairs that encapsulate the core of the material. The complexity lies in the substantial task of extracting pivotal information from the dense fabric of scientific texts, requiring researchers to craft meaningful question-answer pairs that capture the essence of the material.

Current methodologies within this domain often lean on general-purpose language models for information extraction. However, these approaches need help with text refinement and the accurate incorporation of equations. In response, a team of MIT researchers introduced MechGPT, a novel model grounded in a pretrained language model. This innovative approach employs a two-step process, utilizing a general-purpose language model to formulate insightful question-answer pairs. Beyond mere extraction, MechGPT enhances the clarity of key facts.

The journey of MechGPT commences with a meticulous training process implemented in PyTorch within the Hugging Face ecosystem. Based on the Llama 2 transformer architecture, the model flaunts 40 transformer layers and leverages rotary positional embedding to facilitate extended context lengths. Employing a paged 32-bit AdamW optimizer, the training process attains a commendable loss of approximately 0.05. The researchers introduce Low-Rank Adaptation (LoRA) during fine-tuning to augment the model’s capabilities. This involves integrating additional trainable layers while freezing the original pretrained model, preventing the model from erasing its initial knowledge base. The result is heightened memory efficiency and accelerated training throughput.

Year 2021 face_with_colon_three

Senescent cells accumulate in organs during aging, promote tissue dysfunction, and cause numerous aging-related diseases like cancer. The cells arise through a process called “cellular senescence,” a permanent cell cycle arrest resulting from multiple stresses.

A collaborative research group led by Professor Makoto Nakanishi of the Institute of Medical Science, The University of Tokyo (IMSUT), and co-researchers have identified an inhibitor of the glutamate metabolic enzyme GLS1so that its administration selectively eliminates senescent cells in vivo.

Continue reading “GLS1 inhibitor selectively eliminates senescent cells, ameliorates age-associated disorders” »

She gets a new cybernetic arm.

Watch this teen‘s wholesome reaction to getting her very first bionic arm.

IPSC model unveils matrix metalloproteinases’ role in COL4A1/A2 vessel disease.

Sean Carroll speaking at the 6th International FQXi Conference, “Mind Matters: Intelligence and Agency in the Physical World.”

The Foundational Questions Institute (FQXi) catalyzes, supports, and disseminates research on questions at the foundations of physics and cosmology, particularly new frontiers and innovative ideas integral to a deep understanding of reality but unlikely to be supported by conventional funding sources.

Continue reading “Sean Carroll on Causality and the Arrow of Time” »

Investors raced to reinstate Altman after the OpenAI board fired him Friday. But Elon Musk says for AI safety reasons the true reason for the firing must be revealed.

On July 12, 2023, a new research paper was published in Aging, titled, “Chemically induced reprogramming to reverse cellular aging.”

BUFFALO, NY– July 12, 2023 – In a groundbreaking study, researchers have unlocked a new frontier in the fight against aging and age-related diseases. The study, conducted by a team of scientists at Harvard Medical School, has published the first chemical approach to reprogram cells to a younger state. Previously, this was only achievable using a powerful gene therapy.

On July 12, 2023, researchers Jae-Hyun Yang, Christopher A. Petty, Thomas Dixon-McDougall, Maria Vina Lopez, Alexander Tyshkovskiy, Sun Maybury-Lewis, Xiao Tian, Nabilah Ibrahim, Zhili Chen, Patrick T. Griffin, Matthew Arnold, Jien Li, Oswaldo A. Martinez, Alexander Behn, Ryan Rogers-Hammond, Suzanne Angeli, Vadim N. Gladyshev, and David A. Sinclair from Harvard Medical School, University of Maine and Massachusetts Institute of Technology (MIT) published a new research paper in Aging, titled, “Chemically induced reprogramming to reverse cellular aging.”

face_with_colon_three year 2021.

UTokyo People NAKANISH Makoto

UTokyo People JOHMURA Yoshikazu

Continue reading “GLS1 inhibitor that selectively removes senescent cells ameliorated age-associated tissue dysfunction and diseases such as arteriosclerosis” »