Welcome to the October 25, 2023, edition of ACM TechNews, providing timely information for IT professionals three times a week.

ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).

To view "Headlines At A Glance," hit the link labeled "Click here to view this online" found at the top of the page in the html version. The online version now has a button at the top labeled "Show Headlines."
AI Firms Must Be Held Responsible for Harm They Cause, 'Godfathers’ Say
The Guardian (U.K.)
Dan Milmo
October 24, 2023


A group of experts including "godfathers" of artificial intelligence (AI) Geoffrey Hinton and Yoshua Bengio, both ACM Turing Award recipients, said AI companies must be held accountable for the damage their products cause, ahead of an AI safety summit in London. The University of California, Berkeley's Stuart Russell, one of 23 experts who composed AI policy proposals released Tuesday, called developing increasingly powerful AI systems before understanding how to render them safe "utterly reckless." The proposed policies include having governments and companies commit 33% of their AI research and development resources to safe and ethical AI use. Companies that discover dangerous capabilities in their AI models also must adopt specific safeguards.

Full Article
Adaptive Optical Neural Network Connects Thousands of Artificial Neurons
University of Münster (Germany)
October 23, 2023


A team of scientists from Germany's University of Münster and the U.K.'s universities of Exeter and Oxford developed an event-based architecture that enable a photonic processor with adaptive neural connectivity. The researchers used a network of nearly 8,400 optical neurons constructed from waveguide-coupled phase-change material to demonstrate that the strength of the link between two neurons (called a synapse) can be adjusted, while new connections can be formed or existing ones removed. They generated synapses from the respective wavelength and intensity of the optical pulse, facilitating the integration and optical connection of several thousand neurons on one chip.

Full Article

NASA Sends Software Update Over 12 Billion Miles to Voyager 2 NASA Sends Software Update Over 12 Billion Miles to Voyager 2
PC Magazine
Matthew Humphries
October 23, 2023


The U.S. National Aeronautics and Space Administration (NASA) is installing a software patch on the Voyager 2 space probe to avoid an incorrect mode switch that plagued predecessor Voyager 1's attitude articulation and control system (AACS) last year. Voyager project manager Suzanne Dodd described the update as "like an insurance policy that will protect us in the future and help us keep these probes going as long as possible." The update's transmission from Earth across more than 12 billion miles of space took over 18 hours. NASA is reading out AACS memory to ensure the patch does not overwrite essential code or have unintended effects before activation of the update on Oct. 28.

Full Article

The technology uses far-red light to capture clear video of the pupil in dark eye colors. Smartphone Attachment Could Increase Racial Fairness in Neurological Screening
UC San Diego Today
Liezel Labios
October 24, 2023


A smartphone attachment developed by University of California San Diego researchers could be used for low-cost neurological screenings, to ensure accurate results regardless of the user's skin tone. The attachment is placed over the smartphone's camera to take videos and measurements of the subject’s pupil, as changes in pupil size during certain tasks can help assess an individual's neurological functions. The attachment helps conventional color smartphone cameras, which have a hard time distinguishing the pupil from the iris in individuals with dark eye colors. The attachment features a specialized filter that allows far-red light into the camera while blocking other wavelengths, making it easier for the camera to see the pupil because the iris appears much lighter.

Full Article

Staff members and students protest outside the Lipsius building at Leiden University Students, Professors Decry Sensors in Buildings
Nature
Anne Gulland; Fayth Tan
October 20, 2023


Officials at the U.K.'s Queen Mary University of London (QMUL) announced earlier this year that sensors would be installed in campus buildings to assess whether they were being used to their full potential. Although QMUL officials indicated images generated by the sensors would be converted into coordinates that provide real-time data on the number of people in various areas of the buildings so privacy would not be an issue, staff and students remain concerned that the sensors are intended for surveillance. Similar concerns arose in June at the University of California, San Diego, when researchers learned that sensors had been installed in their workplaces as part of the university's Live Density Program. Electronic Frontier Foundation's Jason Kelley said that nonprofit watchdog has found such sensor data "often ends up being used for disciplinary purposes."

Full Article
A Game Changer for Building Robust Distributed Systems
EPFL News (Switzerland)
October 23, 2023


A distributed algorithm developed by researchers at Switzerland's École Polytechnique Fédérale de Lausanne (EPFL) helps overcome performance and reliability issues associated with consensus protocols. The QuePaxa algorithm is an asynchronous consensus protocol that is on par with leader-based protocols. Said EPFL's Bryan Ford, "QuePaxa is just as fast, efficient, low latency, and low cost in terms of network bandwidth, under normal conditions” as widely deployed leader-based protocols. QuePaxa allows a second leader to help the first leader without interference, with the potential for a third leader to assist both of them. Additionally, QuePaxa can survive noisy networks, high communication delays, unpredictably varying network delays, and deliberate denial-of-service attacks.

Full Article
3D-Printed Aluminum Alloy with Unprecedented Fatigue Resistance
City University of Hong Kong
October 24, 2023


Scientists at the City University of Hong Kong and China's Shanghai Jiao Tong University three-dimensionally (3D)-printed an aluminum alloy with enhanced fatigue resistance. The researchers fabricated titanium diboride nanoparticle-decorated (NTD) AlSi10Mg powders into the NTD-Al alloy, whose 260 megapascals of fatigue resistance is more than twice that of other additively manufactured aluminum alloys. Micro-computed tomographic scans revealed the NTD-Al alloy has a continuous 3D-dual-phase cellular nanostructure resists localized damage accumulation and fatigue crack initiation. The scientists said the bulk NTD-Al alloy's fatigue strength exceeded that of all other aluminum alloys, including conventional high-strength wrought alloys.

Full Article

eattle is beating other West Coast tech hubs in attracting talent with cutting-edge skills The U.S. Cities with the Most Cutting-Edge Tech Workers
The Wall Street Journal
Kevin McAllister
October 21, 2023


The Burning Glass Institute's ranking of established and aspiring U.S. tech hubs found that Seattle has the highest proportion of advanced tech workers with cutting-edge skills, which are the most in-demand and well-paid. Burning Glass' Matt Sigelman said, "This ranking has a lot of really important lessons here for cities and some warning signs. You can be a city with a relatively large tech workforce, but if that tech workforce doesn't have the same concentration of the most valuable talent, of the most valuable skills, then you can lose your edge very quickly." Rounding out the top 10 list of cities with the most cutting-edge workers were the San Jose, San Francisco, Boston, Austin, San Diego, New York, Los Angeles, Portland, and Raleigh metropolitan areas.

Full Article
*May Require Paid Registration
Survey Sees Spike in Untested Code Leading to DevOps Crisis
DevOps.com
Mike Vizard
October 20, 2023


In a survey of 500 software developers by market research firm OnePoll, two-thirds of respondents said they pushed untested code into a production environment, with 28% doing so regularly. Sixty percent of respondents to the survey, commissioned by software testing platform Sauce Labs, acknowledged using untested code produced by the ChatGPT large language model, with 26% doing so regularly. Over 66% also acknowledged having merged their own pull requests without a review, with 28% admitting they do so often or very frequently. Fully three-quarters of respondents confessed to bypassing security protocols, with 70% circumventing restrictions for data and/or internal systems access with a coworker's credentials. Sauce Labs' Jason Baum attributed much of this behavior to developers looking for shortcuts because they take on extra work, in addition to citing a shortage of full stack developers able to manage the complete software development lifecycle.

Full Article

new data poisoning tool lets artists fight back against generative AI Data Poisoning Tool Lets Artists Fight Back Against Generative AI
MIT Technology Review
Melissa Heikkilä
October 23, 2023


Researchers led by the University of Chicago's Ben Zhao developed a tool that allows artists to make invisible changes to the pixels in their work before uploading it online, with the goal of damaging future versions of image-generating artificial intelligence (AI) models if the corrupted work is scraped into an AI training set. The researchers plan to integrate the Nightshade data poisoning tool into another tool, Glaze, which lets artists make invisible pixel changes that cause machine learning models to misinterpret the image. Nightshade makes it difficult for AI developers to fix malfunctioning models, as they must locate and remove each corrupted sample. Zhao said thousands of poisoned samples would be necessary to damage the biggest models, which are trained on billions of data samples.

Full Article

Scheme of safety evaluation by DANGER analysis DANGER Analysis Tool for Safer Design of Gene Editing
Hiroshima University (Japan)
October 23, 2023


The Deleterious and ANticipatable Guides Evaluated by RNA-sequencing (DANGER) software tool developed by researchers at Japan's Hiroshima University offers a safer means of genome editing. The researchers designed DANGER to address challenges associated with CRISPR technology, including its inability to quantitatively monitor unexpected phenotypic effects and the method's reliance on a reference genome. They performed risk-averse on- and off-target evaluation in RNA-sequencing data using gene-edited samples of human cells and zebrafish brains. The researchers showed DANGER detected potential DNA on- and off-target sites in the mRNA-transcribed region on the genome, assessing phenotypic effects by deleterious off-target sites according to evidence provided by gene expression changes. The tool also measured gene ontology-term-level phenotypic risk without a reference genome, demonstrating its utility for various organisms, personal human genomes, and disease- and virus-generated atypical genomes.

Full Article

Air pressure within the two channels of the robotic catheter tip Soft Robotic System Aims to Streamline Brain Surgery
National Institute of Biomedical Imaging and Bioengineering
October 20, 2023


Researchers at Johns Hopkins University (JHU) and the University of Maryland developed a soft robotic tool and control system to assist with brain surgery. The researchers designed a three-dimensionally-printed pneumatic catheter tip featuring two parallel channels that cause the tip to deflect left or right when pressurized individually. A hand dial enables precise adjustments of tip position, while haptic feedback signals when the tip is bent; operators can move the catheter forward with one hand while adjusting tip angle with the other. The researchers had a skilled neurosurgeon and a novice use the system to contact an array of five two-millimeter-wide cylindrical targets less than two millimeters apart. JHU's Axel Kreiger said, "Both users were able to use the robotic system for sub-millimeter precision control."

Full Article

The largest quantum computer yet built, created by Atom Computing Quantum Computer Has More Than 1,000 Qubits
New Scientist
Alex Wilkins
October 24, 2023


Quantum computer developer Atom Computing has built a system with a record-breaking 1,180 quantum bits (qubits), more than double the 433-qubit capacity of IBM's Osprey system. The previous largest quantum computers use ultra-cooled superconducting wires as qubits, while Atom Computing's machine uses neutral ytterbium atoms caged by lasers in a two-dimensional grid. The neutral atoms support coherence times of nearly 60 seconds, compared to IBM's Osprey, which only supports about 70 to 80 microseconds of coherence. Atom Computing's Rob Hays said the system is easily scalable, adding that his team intends to multiply the number of qubits in the system by about 10-fold every few years.

Full Article
ACM Transactions on Evolutionary Learning and Optimization
 
ACM Career and Job Center
 

Association for Computing Machinery

1601 Broadway, 10th Floor
New York, NY 10019-7434
1-800-342-6626
(U.S./Canada)



ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: [email protected]