Thomas Fuchs, the Dean of Artificial Intelligence and Human Health at Mount Sinai in NYC, said AI will be needed to retain the standard of care in the U.S. Microsoft is partnering with the digital pathology company Paige to build the world's largest image-based artificial intelligence (AI) model to help detect cancer, the companies announced. The AI model will be used for digital pathology and oncology, configured with billions of parameters to provide a computer vision AI that is orders of magnitude larger than any similar model existing today. Dr. Thomas Fuchs, Paige's founder and chief scientist, told FOX News Digital that the amount of data used in the model is "orders of magnitude" larger than anything made public by Google or Facebook. "It's so much larger than anything that has been published in that area ever," he said. That scale is essential for patients.

Thomas Fuchs, the Dean of Artificial Intelligence and Human Health at Mount Sinai in NYC, said AI will be needed to retain the standard of care in the U.S. Artificial intelligence (AI) is helping physicians to diagnose cancer more accurately at much faster rates and at a lower cost than previously possible, according to a scientist working in computational pathology. Dr. Thomas J. Fuchs, the Dean of Artificial Intelligence and Human Health at Mount Sinai in New York City, also works as the chief scientific officer at Paige AI, a company using AI to detect and treat cancer. The latest study from the company tasked 16 pathologists with the review of 610 whole-slide images prepared at multiple institutions globally. They reviewed the slides once without assistance, and then again with assistance from the Pathology Artificial Intelligence Guidance Engine (Paige AI). When Paige AI was used, diagnostic errors reduced by 70%.

Specular microscopy assessment of the human corneal endothelium (CE) in Fuchs' dystrophy is challenging due to the presence of dark image regions called guttae. This paper proposes a UNet-based segmentation approach that requires minimal post-processing and achieves reliable CE morphometric assessment and guttae identification across all degrees of Fuchs' dystrophy. We cast the segmentation problem as a regression task of the cell and gutta signed distance maps instead of a pixel-level classification task as typically done with UNets. Compared to the conventional UNet classification approach, the distance-map regression approach converges faster in clinically relevant parameters. It also produces morphometric parameters that agree with the manually-segmented ground-truth data, namely the average cell density difference of -41.9 cells/mm2 (95% confidence interval (CI) [-306.2, 222.5]) and the average difference of mean cell area of 14.8 um2 (95% CI [-41.9, 71.5]). These results suggest a promising alternative for CE assessment.

Researchers believe that artificial intelligence has the potential to usher in an era of faster, cheaper and more fruitful drug discovery and development. Over the years, researchers have used AI to analyze troves of biological data, scouring for differences between diseased and healthy cells and using the information to identify potential treatments. More recently, AI has helped predict which chemical compounds are most likely to effectively target SARS-CoV-2. But with AI's potential in drug development comes a slew of ethical pitfalls -- including biases in computer algorithms and the philosophical question of using AI without human mediation. This is where the field of biomedical ethics -- a branch of ethics focused on the philosophical, social and legal issues in the context of medicine and life sciences -- comes in. In mid-March, adjunct Stanford University lecturer Jack Fuchs, PhD, moderated a discussion about the need for clearly articulated principles when guiding the direction of technological advancements, especially AI-enabled drug discovery.

Using both primary and secondary sources, I discuss the role of espionage in early nuclear history. Nuclear weapons are analogous to AI in many ways, so this period may hold lessons for AI governance. Nuclear spies successfully transferred information about the plutonium implosion bomb design and the enrichment of fissile material. Spies were mostly ideologically motivated. Counterintelligence was hampered by its fragmentation across multiple agencies and its inability to be choosy about talent used on the most important military research program in the largest war in human history. Nuclear espionage most likely sped up Soviet nuclear weapons development, but the Soviet Union would have been capable of developing nuclear weapons within a few years without spying. The slight gain in speed due to spying may nevertheless have been strategically significant. Acknowledgements: I am grateful to Matthew Gentzel for supervising this project and Michael Aird, Christina Barta, Daniel Filan, Aaron Gertler, Sidney Hough, Nat Kozak, Jeffery Ohl, and Waqar Zaidi for providing comments. This research was supported by a fellowship from the Stanford Existential Risks Initiative. This post is a short version of the report, x-posted from EA Forum. The full version with additional sections, an appendix, and a bibliography, is available here. The early history of nuclear weapons is in many ways similar to hypothesized future strategic situations involving advanced artificial intelligence (Zaidi and Dafoe 2021, 4). And, in addition to the objective similarity of the situations, the situations may be made more similar by deliberate imitation of the Manhattan Project experience (see this report to the US House Armed Service Committee).

Modelling functions of sets, or equivalently, permutation-invariant functions, is a long-standing challenge in machine learning. Deep Sets is a popular method which is known to be a universal approximator for continuous set functions. We provide a theoretical analysis of Deep Sets which shows that this universal approximation property is only guaranteed if the model's latent space is sufficiently high-dimensional. If the latent space is even one dimension lower than necessary, there exist piecewise-affine functions for which Deep Sets performs no better than a na\"ive constant baseline, as judged by worst-case error. Deep Sets may be viewed as the most efficient incarnation of the Janossy pooling paradigm. We identify this paradigm as encompassing most currently popular set-learning methods. Based on this connection, we discuss the implications of our results for set learning more broadly, and identify some open questions on the universality of Janossy pooling in general.

The Natural Language Inference (NLI) task is an important task in modern NLP, as it asks a broad question to which many other tasks may be reducible: Given a pair of sentences, does the first entail the second? Although the state-of-the-art on current benchmark datasets for NLI are deep learning-based, it is worthwhile to use other techniques to examine the logical structure of the NLI task. We do so by testing how well a machine-oriented controlled natural language (Attempto Controlled English) can be used to parse NLI sentences, and how well automated theorem provers can reason over the resulting formulae. To improve performance, we develop a set of syntactic and semantic transformation rules. We report their performance, and discuss implications for NLI and logic-based NLP.

Hulu might not have the answer to The Great Question, but it's cooking up something tHGttG fans may appreciate... if it does things right. The streaming platform is developing an adaptation of Douglas Adams' beloved classic The Hitchhiker's Guide to the Galaxy, according to Deadline. Apparently, the project will be headed by Carlton Cuse, one of the showrunners behind Lost, Bates Motel and Locke & Key, as well as Jason Fuchs, whose writing credits include Gal Gadot's Wonder Woman. Cuse and Fuchs will write and executive produce the project under ABC Signature, the streaming division of ABC Studios, and Cuse's Genre Arts. Fuchs is reportedly writing the pilot script for what Deadline says is "a modern updating of the classic story," which started as a radio series in 1978.

Pathology artificial intelligence (AI) software developer Paige is highlighting a paper published July 15 in Nature Medicine that indicates the company's technology can be used to develop AI algorithms with "near-perfect accuracy" for analyzing pathology slides for prostate cancer, skin cancer, and breast cancer. In the paper, Chief Scientific Officer Thomas Fuchs, PhD, of Memorial Sloan Kettering Cancer Center and colleagues describe how a series of deep-learning algorithms for clinical decision support in pathology were developed with an automated training and testing technique. Fuchs is the senior author on the paper, with his student Gabriele Campanella as the first author. The deployment of clinical decision support for pathology has been hindered by the need to curate large, manually annotated datasets to test and train AI algorithms, the authors noted. Instead, Campanella et al present a system in which algorithms are trained using only the reported diagnoses.

Pathology laboratories are big data environments. However, these big data are often hidden behind expert humans who manually and with great care visually parse large complex and detailed datasets to provide critical diagnoses. Humans it turns out, are amazingly detailed and accurate large data visualization, segmentation and interpretation devices. Experts are able to zoom in and identify the potentially five or six tumor glands from a large area of stained tissue that comprise the average cancer positive needle biopsy. However, pathology is still an extremely manual and detailed process requiring great skill and accuracy to avoid any potential misdiagnosis.