Academic paper: “Evaluating modern ML models is hard. The strong incentive for researchers and companies to report a state-of-the-art result on some metric often leads to questionable research practices (QRPs): bad practices which fall short of outright research fraud. We describe 43 such practices which can undermine reported results, giving examples where possible. Our list… Continue Reading
Accurate, Focused Research on Law, Technology and Knowledge Discovery Since 2002
Category Archives: AI
AI-powered transcription tool used in hospitals invents things no one ever said
AP: “Tech behemoth OpenAI has touted its artificial intelligence-powered transcription tool Whisper as having near “human level robustness and accuracy.” But Whisper has a major flaw: It is prone to making up chunks of text or even entire sentences, according to interviews with more than a dozen software engineers, developers and academic researchers. Those experts… Continue Reading
“Not Medically Necessary”
ProPublica: Inside the Company Helping America’s Biggest Health Insurers Deny Coverage for Care: “Every day, patients across America crack open envelopes with bad news. Yet another health insurer has decided not to pay for a treatment that their doctor has recommended. Sometimes it’s a no for an MRI for a high school wrestler with a… Continue Reading
DataTalk A Campaign Finance Agent (Beta)
Stanford Open Virtual Assistant Lab – DataTalk – Chat with publicly available campaign finance data from the Federal Election Commission (FEC) and PAC classification data from OpenSecrets.org Continue Reading
VISA Biannual Threat Report Fall 2024
This report provides an overview of the top payments ecosystem threats within the past six-month period (January – June 2024) as identified by Visa Payment Fraud Disruption (PFD). In the December 2023 Biannual Report, Visa PFD noted an interesting shift in threat actors’ organization, access to tools, and target choice, with threat actors increasing in… Continue Reading
The National Security Case for Public AI
Vanderbilt Policy Accelerator: In a recent op-ed in the Washington Post, OpenAI CEO Sam Altman posed a simple but striking question: “Who will control the future of AI?” Altman frames the choice as between two futures: “Will it be one in which the United States and allied nations advance a global AI that spreads the… Continue Reading
Google Scholar is not broken (yet) but there are alternatives
London School of Economics: “…Google Scholar has advantages over traditional academic databases like Scopus and Web of Science: it’s free to use, requires no log in for searching, and has more comprehensive coverage, especially of non-journal sources such as books and theses. These benefits are particularly important for unaffiliated scholars without institutional access to resources,… Continue Reading
Anthropic’s new AI model can control your PC
TechCrunch: “In a pitch to investors last spring, Anthropic said it intended to build AI to power virtual assistants that could perform research, answer emails, and handle other back-office jobs on their own. The company referred to this as a “next-gen algorithm for AI self-teaching” — one it believed that could, if all goes according… Continue Reading
AI Audiobook Narrators in OverDrive and the Issue of Library AI Circulation Policy
STBT: “OverDrive is the company that provides a lot of digital content to libraries. If you’ve borrowed an ebook or an audiobook in Libby, or read a magazine in Kanopy, that’s OverDrive. It seems there is some AI weirdness with audiobook narration on OverDrive, and the narrator is only part of the story. On Monday,… Continue Reading
Google’s AI Fails At 43% Of Finance Queries, Study Finds
Search Engine Journal: A study by The College Investor finds significant inaccuracies in Google’s AI-generated summaries for finance queries. Out of 100 personal finance searches, 43% had misleading or incorrect information. Key Findings -The study evaluated AI overviews across various financial topics, including banking, credit, investing, taxes, and student loans. The results showed: 57% of… Continue Reading
Pete Recommends – Weekly highlights on cyber security issues, October 19, 2024
Via LLRX – Pete Recommends – Weekly highlights on cyber security issues, October 19, 2024 – Privacy and cybersecurity issues impact every aspect of our lives – home, work, travel, education, finance, health and medical records – to name but a few. On a weekly basis Pete Weiss highlights articles and information that focus on the… Continue Reading
AI Detectors Falsely Accuse Students of Cheating With Big Consequences
Bloomberg: “About two-thirds of teachers report regularly using tools for detecting AI-generated content. At that scale, even tiny error rates can add up quickly…Since OpenAI’s ChatGPT brought generative AI to the mainstream almost two years ago, schools have raced to adapt to a changed landscape. Educators now rely on a growing crop of detection tools… Continue Reading
