As multimodal large language models (MLLMs) grow increasingly capable, fixed benchmarks are gradually losing their effectiveness in evaluating high-level scientific understanding. In this paper, we introduce the Multimodal Academic Cover benchmark (MAC), a live benchmark that could continuously evolve with scientific advancement and model progress. MAC leverages over 25,000 image-text pairs sourced from issues of top-tier scientific journals such as Nature, Science, and Cell, challenging MLLMs to reason across abstract visual and textual scientific content. Experiments on our most recent yearly snapshot, MAC-2025, reveal that while MLLMs demonstrate strong perceptual abilities, their cross-modal scientific reasoning remains limited. To bridge this gap, we propose DAD, a lightweight inference-time approach that enhances MLLMs by extending MLLM visual features with language space reasoning, achieving performance improvements of up to 11%. Finally, we highlight the live nature of MAC through experiments on updating journal covers and models for curation, illustrating its potential to remain aligned with the frontier of human knowledge. We release our benchmark at https://github.com/mhjiang0408/MAC_Bench.

Human knowledge is largely implicit and relational -- do we have a friend in common? can I walk from here to there? In this work, we leverage the combinatorial structure of graphs to quantify human priors over such relational data. Our experiments focus on two domains that have been continuously relevant over evolutionary timescales: social interaction and spatial navigation. We find that some features of the inferred priors are remarkably consistent, such as the tendency for sparsity as a function of graph size. Other features are domain-specific, such as the propensity for triadic closure in social interactions. More broadly, our work demonstrates how nonclassical statistical analysis of indirect behavioral experiments can be used to efficiently model latent biases in the data.

My new science fiction/science fact article for Science Robotics is out on why deep ocean robotics is hard. Especially when trying to bring up a sunken submarine 3 miles underwater, which the CIA actually did in 1974. It's even harder if you're trying to bring up an alien spaceship- which is the plot of Harry Turtledove's new sci-fi novel Three Miles Under. Though the expedition was 50 years before the OceanGate Titan tragedy, the same challenges exist for today's robots. The robotics science in the book is very real, the aliens, not so much.

What is unique about AI is also what is most feared and celebrated--its ability to match some of our own skills, and then to go further, accomplishing what humans cannot. AI's capacity to model itself on human behavior has become its defining feature. Yet behind every advance in machine learning and large language models are, in fact, people--both the often obscured human labor that makes large language models safer to use, and the individuals who make critical decisions on when and how to best use this technology. Reporting on people and influence is what TIME does best. That led us to the TIME100 AI.

To help illustrate our cover story on how the AI arms race is changing the world, we reached out to award-winning AI artist Pindar Van Arman, who uses artificial intelligence to create his art. Van Arman, who built his first "painting robot" 15 years ago, uses deep learning neural networks, artificial intelligence, feedback loops, and computational creativity to guide his newer robots. As a result, the robots end up making a surprising number of independent aesthetic decisions in the course of painting each piece--putting a different spin on the idea of "generative" AI: artificial intelligence that doesn't just compute, but also creates. "My machines have grown beyond being simple assistants and are now effectively augmenting my own creativity, as well as having creativity of their own," says Van Arman. "They have become a generative AI art system so sophisticated that it has forced me to consider the possibility that all art is generative."

The global semiconductor industry is getting increasingly crowded as newcomers pour in from all fronts seeking to gain a foothold in advanced chips to power new technologies. The rise of artificial intelligence (AI) has fueled demand for high-capacity chips as tech companies and device makers race to deliver smarter services and products. The global chip shortage throughout 2021 prompted many of them to rely more on themselves.

Australia's extended and disastrous bushfire season has brought into sharp relief the high economic and personal cost of climate change. That economic impact is increasingly recognised around the world as a major business risk. In California, for instance, it led to what is now referred to as the first climate change bankruptcy: the failure of Gas and Electric. The company was brought low by litigation after its equipment was blamed for the Californian wildfires. It is not the only example.

The list of different types of robots which could be used in our daily life is as long as their possible areas of application. Based on media equation assumptions, people are inclined to perceive the robot as an alive social entity. Since it is not common to switch off a social interaction partner, people should be reluctant to switch off the robot they just interacted with, especially when it displays social skills and an autonomous objection against being switched off. To extend previous research as well as media equation findings, the aim of this study is to examine whether an emphatically and rather humanlike behaving robot is perceived as more alive than a machinelike behaving robot and whether this perception influences people's reluctance to switch off the robot. When people are interacting with different media, they often behave as if they were interacting with another person and apply a wide range of social rules mindlessly.According to Researchers, "individuals' interactions with computers, television, and new media are fundamentally social and natural, just like interactions in real life".This phenomenon is described as media equation theory, which stands for "media equal real life".

Cover stories offer a look at the process behind the art on the cover: who made it, how it got made, and why. As described in a Report in this week's issue, an unusual geometric property arises in a bilayer graphene structure in which one layer is rotated 30 degrees with respect to the other. Tiles of triangles, rhombuses, and squares can be mapped over this grid in a pattern called Stampfli tiling. While looking at a figure from this Report (Figure 1), I thought, "No problem. I'll fire up Illustrator, draw the three different shapes, and get to work."

Last week NAB kicked off its latest wave of job cuts billed as part of a restructure to "simplify" the bank. It's part of a three-year plan announced in 2017 to axe 6000 jobs while adding 2000 new technology roles. "As we simplify, we automate processes and things move to digital channels, we will need less people and as that happens we estimate that there will be 6000 less people needed in three years' time," NAB CEO Andrew Thorburn explained when announcing the cuts late last year. "Having said that, we're hiring 2000 people with different capabilities: data scientists, AI, robotics, automation, technology people, digital people, so the net [job loss] will be 4000 and that's just a reshaping that's going to happen." As machines learn how to perform tasks that would otherwise be done by humans, businesses will find themselves needing fewer staff to complete the same amount of work.