Inference-time scaling has emerged as a powerful alternative to parameter scaling for improving language model performance on complex reasoning tasks. While existing methods have shown strong performance gains under fixed compute budgets, there has been little focus on optimally utilizing that budget during inference. In this work, we introduce A*-decoding, a search-based inference-time strategy that builds on the A* search algorithm to optimally utilize a fixed compute budget by prioritizing high-quality reasoning paths during generation. We frame language model decoding as a structured search in a state space of partial solutions, applying the A* transition model to identify promising continuations guided by an external process supervision signal. In our experiments, A*-decoding reaches the performance levels of strong inference scaling baselines like best-of-N and particle filtering while using up to 3x fewer tokens and 30% fewer PRM passes under equivalent compute budgets. On the MATH500 and AIME 2024 benchmarks, A*-decoding enables Llama-3.2-1B-Instruct to match the performance of the 70x larger Llama-3.1-70B-Instruct, and allows Qwen3-1.7B to reach o1-like reasoning accuracy. These results highlight the power of structured search in decoding, offering an alternative to brute-force sampling or scale-driven gains. Our work demonstrates how thoughtful inference-time strategies can enhance reasoning in SLMs, pointing toward future advances in more efficient and scalable language model deployment.

--Cross-view geo-localisation identifies coarse geographical position of an automated vehicle by matching a ground-level image to a geo-tagged satellite image from a database. Despite the advancements in Cross-view geo-localisation, significant challenges still persist such as similar looking scenes which makes it challenging to find the correct match as the top match. Existing approaches reach high recall rates but they still fail to rank the correct image as the top match. T o address this challenge, this paper proposes GeoVLM, a novel approach which uses the zero-shot capabilities of vision language models to enable cross-view geo-localisation using interpretable cross-view language descriptions. GeoVLM is a trainable reranking approach which improves the best match accuracy of cross-view geo-localisation. GeoVLM is evaluated on standard benchmark VIGOR and University-1652 and also through real-life driving environments using Cross-View United Kingdom, a new benchmark dataset introduced in this paper . The results of the paper show that GeoVLM improves retrieval performance of cross-view geo-localisation compared to the state-of-the-art methods with the help of explainable natural language descriptions. The code is available at https://github.com/CA V-Research-Lab/GeoVLM Index T erms --cross-view geo-localisation, automated vehicles, vision-language models, satellite imagery, interpretable AI, image retrieval. OCALISA TION in automated vehicles refer to the process of finding the precise position and orientation of the automated system or a robot within a given environment relative to a chosen reference coordinate system [1]. Localisation in automated vehicles serves as a backbone for higher-level functions such as perception, planning, and control, ensuring the vehicle can navigate safely and effectively. The most common solution for estimating the geo-position of automated vehicles is Global Positioning System (GPS).

Controlled synthesis of materials with specified atomic structures underpins technological advances yet remains reliant on iterative, trial-and-error approaches. Nanoparticles (NPs), whose atomic arrangement dictates their emergent properties, are particularly challenging to synthesise due to numerous tunable parameters. Here, we introduce an autonomous approach explicitly targeting synthesis of atomic-scale structures. Our method autonomously designs synthesis protocols by matching real time experimental total scattering (TS) and pair distribution function (PDF) data to simulated target patterns, without requiring prior synthesis knowledge. We demonstrate this capability at a synchrotron, successfully synthesising two structurally distinct gold NPs: 5 nm decahedral and 10 nm face-centred cubic structures. Ultimately, specifying a simulated target scattering pattern, thus representing a bespoke atomic structure, and obtaining both the synthesised material and its reproducible synthesis protocol on demand may revolutionise materials design. Thus, ScatterLab provides a generalisable blueprint for autonomous, atomic structure-targeted synthesis across diverse systems and applications.

The AI data center also promises to transform the state's energy future. Stretching in length for more than a mile, it will be Meta's largest in the world, and it will have an enormous appetite for electricity, requiring two gigawatts for computation alone (the electricity for cooling and other building needs will add to that). When it's up and running, it will be the equivalent of suddenly adding a decent-size city to the region's grid--one that never sleeps and needs a steady, uninterrupted flow of electricity. To power the data center, Entergy aims to spend 3.2 billion to build three large natural-gas power plants with a total capacity of 2.3 gigawatts and upgrade the grid to accommodate the huge jump in anticipated demand. In its filing to the state's power regulatory agency, Entergy acknowledged that natural-gas plants "emit significant amounts of CO2" but said the energy source was the only affordable choice given the need to quickly meet the 24-7 electricity demand from the huge data center.

This story is a part of MIT Technology Review's series "Power Hungry: AI and our energy future," on the energy demands and carbon costs of the artificial-intelligence revolution. These somewhat unlikely partnerships could be a win for both the nuclear power industry and large tech companies. Tech giants need guaranteed sources of energy, and many are looking for low-emissions ones to hit their climate goals. For nuclear plant operators and nuclear technology developers, the financial support of massive established customers could help keep old nuclear power plants open and push new technologies forward. "There [are] a lot of advantages to nuclear," says Michael Terrell, senior director of clean energy and carbon reduction at Google.

"Dollars are being invested, GPUs are being burned, water is being evaporated--it's just absolutely the wrong direction," says Ali Farhadi, CEO of the Seattle-based nonprofit Allen Institute for AI. But sift through the talk of rocketing costs--and climate impact--and you'll find reasons to be hopeful. There are innovations underway that could improve the efficiency of the software behind AI models, the computer chips those models run on, and the data centers where those chips hum around the clock. Here's what you need to know about how energy use, and therefore carbon emissions, could be cut across all three of those domains, plus an added argument for cautious optimism: There are reasons to believe that the underlying business realities will ultimately bend toward more energy-efficient AI. The most obvious place to start is with the models themselves--the way they're created and the way they're run.

Despite the fact that billions of dollars are being poured into reshaping energy infrastructure around the needs of AI, no one has settled on a way to quantify AI's energy usage. Worse, companies are generally unwilling to disclose their own piece of the puzzle. There are also limitations to estimating the emissions associated with that energy demand, because the grid hosts a complicated, ever-changing mix of energy sources. So, that said, here are the many variables, assumptions, and caveats that we used to calculate the consequences of an AI query. Companies like OpenAI, dealing in "closed-source" models, generally offer access to their systems through an interface where you input a question and receive an answer.

AI's integration into our lives is the most significant shift in online life in more than a decade. Hundreds of millions of people now regularly turn to chatbots for help with homework, research, coding, or to create images and videos. Today, new analysis by MIT Technology Review provides an unprecedented and comprehensive look at how much energy the AI industry uses--down to a single query--to trace where its carbon footprint stands now, and where it's headed, as AI barrels towards billions of daily users. This story is a part of MIT Technology Review's series "Power Hungry: AI and our energy future," on the energy demands and carbon costs of the artificial-intelligence revolution. We spoke to two dozen experts measuring AI's energy demands, evaluated different AI models and prompts, pored over hundreds of pages of projections and reports, and questioned top AI model makers about their plans.

Innovation in IT got us to this point. Graphics processing units (GPUs) that power the computing behind AI have fallen in cost by 99% since 2006. There was similar concern about the energy use of data centers in the early 2010s, with wild projections of growth in electricity demand. But gains in computing power and energy efficiency not only proved these projections wrong but enabled a 550% increase in global computing capability from 2010 to 2018 with only minimal increases in energy use. In the late 2010s, however, the trends that had saved us began to break.

Recent advances in large language models (LLMs) have fueled the vision of automated scientific discovery, often called AI Co-Scientists. To date, prior work casts these systems as generative co-authors responsible for crafting hypotheses, synthesizing code, or drafting manuscripts. In this work, we explore a complementary application: using LLMs as verifiers to automate the \textbf{academic verification of scientific manuscripts}. To that end, we introduce SPOT, a dataset of 83 published papers paired with 91 errors significant enough to prompt errata or retraction, cross-validated with actual authors and human annotators. Evaluating state-of-the-art LLMs on SPOT, we find that none surpasses 21.1\% recall or 6.1\% precision (o3 achieves the best scores, with all others near zero). Furthermore, confidence estimates are uniformly low, and across eight independent runs, models rarely rediscover the same errors, undermining their reliability. Finally, qualitative analysis with domain experts reveals that even the strongest models make mistakes resembling student-level misconceptions derived from misunderstandings. These findings highlight the substantial gap between current LLM capabilities and the requirements for dependable AI-assisted academic verification.