Artificial neural networks, prevalent models in machine learning capable of being trained for various tasks, derive their name from their structural resemblance to the information-processing methods…
Everyone knows that arithmetic is true: 2 + 2 = 4.
But surprisingly, we don’t know why it’s true.
By stepping outside the box of our usual way of thinking about numbers, my colleagues and I have recently shown that arithmetic has biological roots and is a natural consequence of how perception of the world around us is organised.
The John Templeton Foundation recently invited biologist Michael Levin to speak to a small group about the presence of agency and cognition in the most fundamental forms of life, even at the levels of cells and tissues. In the recorded video, Dr. Levin, who directs a developmental biology lab at Tufts University, discusses with Philip Ball, a science writer and author of the newly published Book of Minds: How to Understand Ourselves and Other Beings.
Founded in 1987, the John Templeton Foundation supports research and dialogue on the deepest and most perplexing questions facing humankind. The Foundation funds work on subjects ranging from black holes and evolution to creativity, forgiveness, and free will. It also encourages civil, informed dialogue among scientists, philosophers, theologians, and the public at large.
Continue reading “Can Cells Think? The Magic of Developmental Biology” »
Remember when eggs were so high? A vaccine for birds, now that can make money. 🤔
In the past two years, a viral disease has swept across much of the planet — not Covid but a type of avian flu. It’s devastated the poultry industry in the US, Europe, and elsewhere, sickening millions of farmed birds, which either die from infection or are killed by farmers seeking to stem the spread.
The ongoing outbreak of avian flu has killed hundreds of thousands — if not millions — of wild birds, including endangered species like the California condor. It’s one of the worst wildlife disease outbreaks in history. Having now spread across five continents and hundreds of wildlife species, scientists call the current outbreak a panzootic, meaning a pandemic among animals.
Continue reading “A frightening virus is killing a massive number of wild birds” »
The origin and early evolution of life is generally studied under two different paradigms: bottom up and top down. Prebiotic chemistry and early Earth geochemistry allow researchers to explore possible origin of life scenarios. But for these “bottom–up” approaches, even successful experiments only amount to a proof of principle. On the other hand, “top–down” research on early evolutionary history is able to provide a historical account about ancient organisms, but is unable to investigate stages that occurred during and just after the origin of life. Here, we consider ancient electron transport chains (ETCs) as a potential bridge between early evolutionary history and a protocellular stage that preceded it. Current phylogenetic evidence suggests that ancestors of several extant ETC components were present at least as late as the last universal common ancestor of life. In addition, recent experiments have shown that some aspects of modern ETCs can be replicated by minerals, protocells, or organic cofactors in the absence of biological proteins. Here, we discuss the diversity of ETCs and other forms of chemiosmotic energy conservation, describe current work on the early evolution of membrane bioenergetics, and advocate for several lines of research to enhance this understanding by pairing top–down and bottom–up approaches.
Blog post with audio player, show notes, and transcript: https://www.preposterousuniverse.com/podcast/2020/01/06/78-d…fic-image/
Patreon: https://www.patreon.com/seanmcarroll.
Wilfrid Sellars described the task of philosophy as explaining how things, in the broadest sense of term, hang together, in the broadest sense of the term. (Substitute “exploring” for “explaining” and you’d have a good mission statement for the Mindscape podcast.) Few modern thinkers have pursued this goal more energetically, creatively, and entertainingly than Daniel Dennett. One of the most respected philosophers of our time, Dennett’s work has ranged over topics such as consciousness, artificial intelligence, metaphysics, free will, evolutionary biology, epistemology, and naturalism, always with an eye on our best scientific understanding of the phenomenon in question. His thinking in these areas is exceptionally lucid, and he has the rare ability to express his ideas in ways that non-specialists can find accessible and compelling. We talked about all of them, in a wide-ranging and wonderfully enjoyable conversation.
Continue reading “Mindscape 78 | Daniel Dennett on Minds, Patterns, and the Scientific Image” »
The ongoing AI revolution, set to reshape lifestyles and workplaces, has seen deep neural networks (DNNs) play a pivotal role, notably with the emergence of foundation models and generative AI. Yet, the conventional digital computing frameworks that host these models hinder their potential performance and energy efficiency. While AI-specific hardware has emerged, many designs separate memory and processing units, resulting in data shuffling and reduced efficiency.
IBM Research has pursued innovative ways to reimagine AI computation, leading to the concept of analog in-memory computing, or analog AI. This approach draws inspiration from neural networks in biological brains, where synapse strength governs neuron communication. Analog AI employs nanoscale resistive devices like Phase-change memory (PCM) to store synaptic weights as conductance values. PCM devices transition between amorphous and crystalline states, encoding a range of values and enabling local storage of weights with non-volatility.
A significant stride towards making analog AI a reality has been achieved by IBM Research in a recent Nature Electronics publication. They introduced a cutting-edge mixed-signal analog AI chip tailored for various DNN inference tasks. This chip, fabricated at IBM’s Albany NanoTech Complex, features 64 analog in-memory compute cores, each housing a 256-by-256 crossbar array of synaptic unit cells. Integrated compact, time-based analog-to-digital converters facilitate seamless transitions between analog and digital domains. Moreover, digital processing units within each core handle basic neuronal activation functions and scaling operations.
Daniel Dennett discusses the nature of consciousness, if consciousness is an illusion, artificial intelligence and virtual immortality, and how he covers all of this in his book, Just Deserts: Debating Free Will, co-authored with Gregg D. Caruso.
Just Deserts: Debating Free Will https://www.amazon.com/Just-Deserts-Debating-Free-Will/dp/15…atfound-20
Read an excerpt https://www.closertotruth.com/articles/book-excerpt-just-deserts.
Researchers have developed an easy-to-build, low-cost 3D nanoprinting system that can create arbitrary 3D structures with extremely fine features. The new 3D nanoprinting technique is precise enough to print metamaterials as well as a variety of optical devices and components such as microlenses, micro-optical devices and metamaterials.
“Our system uses a two-step absorption process to realize 3D printing with accuracy reaching the nanometer level, which is suitable for commercial manufacturing,” said research team leader Cuifang Kuang from the Zhejiang Lab and Zhejiang University, both in China. “It can be used for a variety of applications such as printing micro or nanostructures for studying biological cells or fabricating the specialized optical waveguides used for virtual and augmented reality devices.”
Conventional high-resolution 3D nanoprinting approaches use pulsed femtosecond lasers that cost tens of thousands of dollars. In Optics Letters, Kuang and colleagues describe their new system based on an integrated fiber-coupled continuous-wave laser diode that is not only inexpensive but also easy to operate.
