"Global AI Cultures," by @Simone_Natale (@unito) et al., says a global approach to AI can only emerge by emphasizing the centrality of #culture, i.e., the #ideas, customs, and social behaviors of a #people, period, group, or #humanity in general.

Cognizant of their impact, #researchers are developing ways to #define, detect, and quantify the effects of code #hallucinations, writes Karen Emslie / @DamnRebelBitch in "Nonsense and Malicious Packages: LLM Hallucinations in Code Generation."

"Scaling Deep Learning Education in a College of Arts & Sciences," by Ambuj Tewari, describes how @UMich created a course to teach #DeepLearning to @UMichLSA students who aren't #ComputerScience or #engineering majors, with guidance for other educators.

"Scrutinizing LLM Reasoning Models," by Emma Stamm / @14floating, looks at the work of #researchers on the steps behind #LLMs' inferences.

"Why Computer Science Is No Good, Redux," by Micah D. Beck (@UTKnoxville), says simplifying assumptions that underpin formal models can ignore the variety of #environmental and operational variables that can alter the behavior of #digital systems.

"ACM Publications Finances for 2023 and 2024," by Scott Delman (@TheOfficialACM), @DameWendyDBE, and Divesh Srivastava, highlights the growth of the ACM Open model and the association's transition to 100% open access publication.

"AI for Senior Citizens," by Sandrine Ceurstemont / @purplesasa, describes how #ArtificialIntelligence can help meet the #healthcare needs of an #aging #population.

"Differentiable Economics: Strategic Behavior, Mechanisms, and Machine Learning," by @Mar_Bichler (@TU_Muenchen) and David C. Parkes (@hseas), describes how recent advances have made #MachineLearning a tool for solving longstanding #problems in #economics.

"AI and the Democratization of Cybercrime," by Alex Williams, shows how tech is shaping a low-barrier #cybercrime ecosystem, and says offensive AI should be regulated like a controlled substance.

"Putting Carbon on the Clock," by Samuel Greengard, describes carbon-aware scheduling tools that assist #datacenter and #tech #energy deployments with an eye on cost, #performance, and #CarbonEmissions.

"Deconstructing the Take It Down Act," by James @Grimmelm (@CornellLaw, @Cornell_Tech), says it's easy to imagine the @FTC weaponizing its newfound #TakeItDownAct authority as a tool of #censorship and #extortion.

"Achieving Early Wins in Generative AI," by Shanmugam Sudalaimuthu, presents a tactical approach to using Structured Outputs to integrate #GenerativeAI with existing IT systems.

"Stop Using Vulnerability Counts to Measure Software Security," by @AndyMeneely (@RIT) and @Nodnarb_Rellek (#RIT), identifies problems with the National #Vulnerability Database and proposes a new way to report #security metrics.

"Visual Object Tracking Challenge: How to Offer a Working Solution in 24 Hours Using Multimodal LLM," by Aleksandr Romanov (@HSE_eng) & Roman Solovyev, describe a video of a black and white cat and some playful problems at a #computer #vision #competition.

"SRAM Has No Chill: Exploiting Power Domain Separation to Steal On-Chip Secrets," by @Jubayer0175 (@AWScloud) and Matthew Hicks (@Virginia_Tech), shows that current on-chip SRAM is resilient against conventional temperature “freezing”-based attacks.

"Understanding Mobile App Reviews to Guide Misuse Audits," by Vaibhav Garg (@Virginia_Tech), Hui Guo (@Quora), Nirav Ajmeri (@BristolUni), Saikath Bhattacharya (@IllinoisStateU, Munindar Singh (@NCState), says app reviews can help identify exploitable apps

"Benchmarks for AI in Software Engineering," by Satish Chandra (@Google), calls the lack of commonly adopted benchmarks for significant portions of the #SoftwareDevelopment lifecycle a missed opportunity for #developers and ML engineers.

Lenore D. Zuck (@thisisUIC), co-author of "It Takes a Village" (, says collaboration between network #developers and the formal methods community would increase the security of nextgen #networking systems, in an original #video,

"It Takes a Village: Bridging the Gaps between Current and Formal Specifications for Protocols," David Basin et al. says the #networking community could benefit from adopting formal methods,

"On Notifications," by @VGCerf (@Google), describes new ideas that are emerging to make #communication among interacting agents more reliable and precise.