The United States under President Donald Trump is ramping up use of surveillance systems and artificial intelligence (AI) to track and arrest immigrants, raising fears that risks to accuracy and privacy could put almost anyone in danger of getting caught up in the crackdown. The Department of Homeland Security (DHS) and other immigration control agencies are using a suite of AI tools -- such as facial recognition scanners in public areas and robotic dogs patrolling the southern border for human movement -- as part of the crackdown on alleged illegal immigration. Many of the AI tools that immigration agents are using have been in place for years and are a legacy of previous administrations, according to Saira Hussain, an attorney at the Electronic Frontier Foundation, a digital rights advocacy group.

Step by mechanical step, dozens of humanoid robots took to the streets of Beijing early Saturday, joining thousands of their flesh-and-blood counterparts in a world-first half-marathon showcasing China's drive to lead the global race in cutting-edge technology. The 21-kilometer event held in the Chinese capital's E-Town -- a state-backed hub for high-tech manufacturing -- is billed as a groundbreaking effort to test the limits of bipedal robots in real-world conditions. At the crack of the starter's gun, and as the Chinese pop song "I Believe" blared out from loudspeakers on repeat, the robots queued up one by one and took their first tentative steps.

And Henk van Ess, a global expert in using AI in investigative research, has proven how useful it can be - but it would be safe to say he does not believe it lies in starter packs. "It's like watching a supercomputer calculate how many Hobnobs fit in a Sports Direct mug, while solving climate change sits on the'to-do' list," he said. But it's the technological equivalent of using the Large Hadron Collider to heat up your Pot Noodle. "While everyone's busy generating these digital equivalents of small talk, they're missing the actually revolutionary stuff AI can do - it's just wasteful to put all that energy into creating digital fluff when we can use it for solving real-world problems."

In order to reduce the cost of experimental evaluation for models, we introduce a computational theory of evaluation for prediction and decision models: build evaluation model to accelerate the evaluation procedures. We prove upper bounds of generalized error and generalized causal effect error of given evaluation models. We also prove efficiency, and consistency to estimated causal effect from deployed subject to evaluation metric by prediction. To learn evaluation models, we propose a meta-learner to handle heterogeneous evaluation subjects space problem. Comparing with existed evaluation approaches, our (conditional) evaluation model reduced 24.1\%-99.0\% evaluation errors across 12 scenes, including individual medicine, scientific simulation, social experiment, business activity, and quantum trade. The evaluation time is reduced 3-7 order of magnitude comparing with experiments or simulations.

Temporal Point Processes (TPPs) are widely used for modeling event sequences in various medical domains, such as disease onset prediction, progression analysis, and clinical decision support. Although TPPs effectively capture temporal dynamics, their lack of interpretability remains a critical challenge. Recent advancements have introduced interpretable TPPs. However, these methods fail to incorporate numerical features, thereby limiting their ability to generate precise predictions. To address this issue, we propose Hybrid-Rule Temporal Point Processes (HRTPP), a novel framework that integrates temporal logic rules with numerical features, improving both interpretability and predictive accuracy in event modeling. HRTPP comprises three key components: basic intensity for intrinsic event likelihood, rule-based intensity for structured temporal dependencies, and numerical feature intensity for dynamic probability modulation. To effectively discover valid rules, we introduce a two-phase rule mining strategy with Bayesian optimization. To evaluate our method, we establish a multi-criteria assessment framework, incorporating rule validity, model fitting, and temporal predictive accuracy. Experimental results on real-world medical datasets demonstrate that HRTPP outperforms state-of-the-art interpretable TPPs in terms of predictive performance and clinical interpretability. In case studies, the rules extracted by HRTPP explain the disease progression, offering valuable contributions to medical diagnosis.

Health and Human Services Secretary Robert F. Kennedy Jr. conducts a news conference to discuss the Centers for Disease Control and Prevention's latest Autism and Developmental Disabilities Monitoring Network survey.Tom Williams/CQ Roll Call/AP While his anti-vaccine allies swooned and scientists cringed, HHS Secretary Robert F. Kennedy Jr. used his first-ever press conference this week, in response to new data showing an apparent increase in the number of autistic kids, to promote a variety of debunked, half-true, and deeply ableist ideas about autism. He painted the condition as a terrifying "disease" that "destroys," as he put it, children and their families. Kennedy made it clear he planned to use his powerful role as the person in charge of a massive federal agency devoted to protecting public health to promote the idea that autism is caused by "environmental factors," a still-speculative thesis that's clearly a short walk towards advancing his real aim: blaming vaccines. Kennedy has spent the last 20 years promoting anti-vaccine rhetoric, falsely and repeatedly claiming that vaccines are linked to autism. Yet as the press conference made clear, Kennedy knows startlingly little about autism.

Under no circumstances should you let AI do your schoolwork for you, but Google has decided to make that option a little bit easier for the next year. The company is offering a free year of it's Google One AI Premium plan, which includes Gemini Advanced, access to the AI assistant in the Google Workspace and things like Gemini Live, to any college student willing to sign up. The offer gives you a sample platter of Google's latest AI features, which normally costs 20 per month, and is primarily focused on things you can do with Gemini. That includes experimental products like NotebookLM for analyzing documents, and Whisk for remixing images and videos. Because this is a Google One plan, you'll also get 2TB of Google Drive storage for the parade of PDFs that make up college life.

Sign up for our daily newsletter to get the best of The New Yorker in your in-box. Elon Musk, who's taking his chainsaw to the federal government, is not merely a chaos agent, as he is sometimes described. Jill Lepore, the best-selling author of "These Truths" and other books, says that Musk is animated by obsessions and a sense of mission he acquired through reading, and misreading, science fiction. "When he keeps saying, you know, 'We're at a fork in the road. The future of human civilization depends on this election,' he means SpaceX," she tells David Remnick.

The message from experts is clear: artificial intelligence (AI) can help professionals spend less time on repetitive, time-consuming tasks and more on value-adding activities. However, there's just one issue: what are these value-adding activities? Senior executives may like the sound of highly paid staff spending less time on mundane activities, but only if these professionals use their saved time effectively. So, how can you make the most of the time you save by using AI? Five experts share their top tips. Gabriela Vogel, vice president analyst in Gartner's Executive Leadership of Digital Business practice, said it's crucial to consider how professionals will use spare time in an age of AI -- and it's an area she's researching right now.

Earlier this week via blog post, OpenAI released their newest AI models: o3 and o4-mini. These models are the company's "smartest and most capable models to date" and their first reasoning models that can also reason when it comes to images. In short, these AI models can use an image--such as a photograph or a sketch--as part of an analysis. The models can also adjust, zoom in on, and rotate an image during reasoning. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation.