Classical demand modeling analyzes travel behavior using only low-dimensional numeric data (i.e. sociodemographics and travel attributes) but not high-dimensional urban imagery. However, travel behavior depends on the factors represented by both numeric data and urban imagery, thus necessitating a synergetic framework to combine them. This study creates a theoretical framework of deep hybrid models with a crossing structure consisting of a mixing operator and a behavioral predictor, thus integrating the numeric and imagery data into a latent space. Empirically, this framework is applied to analyze travel mode choice using the MyDailyTravel Survey from Chicago as the numeric inputs and the satellite images as the imagery inputs. We found that deep hybrid models outperform both the traditional demand models and the recent deep learning in predicting the aggregate and disaggregate travel behavior with our supervision-as-mixing design. The latent space in deep hybrid models can be interpreted, because it reveals meaningful spatial and social patterns. The deep hybrid models can also generate new urban images that do not exist in reality and interpret them with economic theory, such as computing substitution patterns and social welfare changes. Overall, the deep hybrid models demonstrate the complementarity between the low-dimensional numeric and high-dimensional imagery data and between the traditional demand modeling and recent deep learning. It generalizes the latent classes and variables in classical hybrid demand models to a latent space, and leverages the computational power of deep learning for imagery while retaining the economic interpretability on the microeconomics foundation.

Membership inference (MI) attack is currently the most popular test for measuring privacy leakage in machine learning models. Given a machine learning model, a data point and some auxiliary information, the goal of an MI attack is to determine whether the data point was used to train the model. In this work, we study the reliability of membership inference attacks in practice. Specifically, we show that a model owner can plausibly refute the result of a membership inference test on a data point $x$ by constructing a proof of repudiation that proves that the model was trained without $x$. We design efficient algorithms to construct proofs of repudiation for all data points of the training dataset. Our empirical evaluation demonstrates the practical feasibility of our algorithm by constructing proofs of repudiation for popular machine learning models on MNIST and CIFAR-10. Consequently, our results call for a re-evaluation of the implications of membership inference attacks in practice.

The House Oversight Committee is stepping up oversight of the Biden administration with eight expected hearings being held over the next three days. "We have eight hearings this week!" The first hearings kick off Wednesday on COVID origins, advances in artificial intelligence, the border crisis and the depletion of the Strategic Petroleum Reserve (SPR). The border hearing will feature testimony from chief border patrol agents and the COVID origins investigation will showcase the former Centers for Disease Control and Prevention (CDC) director and health and science experts. Thursday, the committee will examine the role of the Office of Personnel Management (OPM), the government's largest employer, and also "waste, fraud and abuse" in pandemic spending.

The rise of artificial intelligence is set to boost economic growth, but it is also poised to take over the job market - and a new study reveals the 20 most occupations at risk. A team of researchers led by Princeton University conducted an AI occupational exposure methodology by linking 10 AI-powered applications, such as language modeling, to 52 human abilities to understand if any closely relate. The results showed that telemarketers, teachers, school psychologists and judges are among the highest at risk. Fears of software eliminating human jobs have recently made waves across the globe following the launch of ChatGPT and its ability to perform eerily-human professional tasks such as writing emails and resumes. 'The effect of AI on work will likely be multi-faceted.

Why not more coverage of poverty in America, gross income inequality, the quality of American health care, the lack of decent housing that Americans can actually afford? Regulation Can Be Puzzle to Lawmakers" (front page, March 4): The recent coverage of Washington's response to artificial intelligence is a welcome shift toward an overdue policy debate. But the challenge ahead is not so much about educating lawmakers about new technology -- technologies are always changing -- as it is about establishing the necessary safeguards to protect the public. At the Center for A.I. and Digital Policy, we have closely examined A.I. policies and practices around the world. Europe has taken the lead with the proposed European Union A.I. act.

Individuals should be treated as individuals and not on the basis of their membership in racial groups, especially by our government. Unfortunately, a new executive order encourages federal agencies to focus on racial group identity rather than the character and qualifications of employees and contractors. It will result in racial quotas in hiring, procuring, and even using artificial intelligence throughout the government. The executive order's stated goal is advancing racial equity throughout the federal government. The word "equity" appears 21 times.

He spent about six hours at the property, which was the scene of a quadruple homicide in November. As the University of Idaho community reels from the shocking slayings of four undergrad students in an off-campus rental home in Moscow, Idaho, this past November, school officials have already announced plans to tear the building down. "The owner of the King Street house offered to give the house to the university, which we accepted," University of Idaho President Scott Green said last week. "The house will be demolished. This is a healing step and removes the physical structure where the crime that shook our community was committed."

Vasyl Bilous's last name means "white mustache." His actual mustache is dark brown with a hint of gray. He's worn one since high school. In a picture that he took on the first day of Russia's full-scale invasion of Ukraine, Vasyl has a chevron mustache, a neat barbershop cut--close on the sides, paintbrush-thick on top. At the time, he was an assistant professor of forensics at the National Law University, in Kharkiv, and a lawyer in private practice.

Yesterday, hundreds in Eckersley's community of friends and colleagues packed the pews for an unusual sort of memorial service at the church-like sanctuary of the Internet Archive in San Francisco--a symposium with a series of talks devoted not just to remembrances of Eckersley as a person but a tour of his life's work. Facing a shrine to Eckersley at the back of the hall filled with his writings, his beloved road bike, and some samples of his Victorian goth punk wardrobe, Turan, Gallagher, and 10 other speakers gave presentations about Eckersley's long list of contributions: his years pushing Silicon Valley towards better privacy-preserving technologies, his co-founding of a groundbreaking project to encrypt the entire web, and his late-life pivot to improving the safety and ethics of AI. The event also served as a kind of soft launch for AOI, the organization that will now carry on Eckersley's work after his death. Eckersley envisioned the institute as an incubator and applied laboratory that would work with major AI labs to that take on the problem Eckersley had come to believe was, perhaps, even more important than the privacy and cybersecurity work to which he'd devoted decades of his career: redirecting the future of artificial intelligence away from the forces causing suffering in the world, toward what he described as "human flourishing." "We need to make AI not just who we are, but what we aspire to be," Turan said in his speech at the memorial event, after playing a recording of the phone call in which Eckersley had recruited him.

The large tech companies – Google, Meta/Facebook, Microsoft – are in a race to introduce new artificial intelligence systems and what are called chatbots, that you can have conversations with and are more sophisticated than Siri or Alexa. Microsoft's AI search engine and chatbot, Bing, can be used on a computer or cell phone to help with planning a trip or composing a letter. It was introduced on February 7 to a limited number of people as a test – and initially got rave reviews. But then several news organizations began reporting on a disturbing so-called "alter ego" within Bing Chat, called Sydney. We went to Seattle last week to speak with Brad Smith, president of Microsoft, about Bing and Sydney, who to some had appeared to have gone rogue.