The automated generation of layouts is vital for embodied intelligence and autonomous systems, supporting applications from virtual environment construction to home robot deployment. Current approaches, however, suffer from spatial hallucination and struggle with balancing semantic fidelity and physical plausibility, often producing layouts with deficits such as floating or overlapping objects and misaligned stacking relation. In this paper, we propose AutoLayout, a fully automated method that integrates a closed-loop self-validation process within a dual-system framework. Specifically, a slow system harnesses detailed reasoning with a Reasoning-Reflection-Generation (RRG) pipeline to extract object attributes and spatial constraints. Then, a fast system generates discrete coordinate sets and a topological relation set that are jointly validated. To mitigate the limitations of handcrafted rules, we further introduce an LLM-based Adaptive Relation Library (ARL) for generating and evaluating layouts. Through the implementation of Slow-Fast Collaborative Reasoning, the AutoLayout efficiently generates layouts after thorough deliberation, effectively mitigating spatial hallucination. Its self-validation mechanism establishes a closed-loop process that iteratively corrects potential errors, achieving a balance between physical stability and semantic consistency. The effectiveness of AutoLayout was validated across 8 distinct scenarios, where it demonstrated a significant 10.1% improvement over SOTA methods in terms of physical plausibility, semantic consistency, and functional completeness.

The study explores the capabilities of OpenAI's ChatGPT in solving different types of physics problems. ChatGPT (with GPT-4) was queried to solve a total of 40 problems from a college-level engineering physics course. These problems ranged from well-specified problems, where all data required for solving the problem was provided, to under-specified, real-world problems where not all necessary data were given. Our findings show that ChatGPT could successfully solve 62.5% of the well-specified problems, but its accuracy drops to 8.3% for under-specified problems. Analysis of the model's incorrect solutions revealed three distinct failure modes: 1) failure to construct accurate models of the physical world, 2) failure to make reasonable assumptions about missing data, and 3) calculation errors. The study offers implications for how to leverage LLM-augmented instructional materials to enhance STEM education. The insights also contribute to the broader discourse on AI's strengths and limitations, serving both educators aiming to leverage the technology and researchers investigating human-AI collaboration frameworks for problem-solving and decision-making.

How does your brain innovate? How does the human brain learn to innovate? The Thousand Brain theory, proposed by Silicon Valley innovator-turned-scientist Jeff Hawkins, is the first proposal to explain how the brain functions at the cellular level. This theory has direct, important, unexpected implications for how people learn to innovate, and therefore for how teachers, coaches, managers, and other professionals teach innovation. This article provides a summary of Jeff's conclusions.

Christopher Intagliata: Artificial intelligence systems have bested humans at chess, poker, Jeopardy, Go, and countless other games. Susan Hespos: They still can't do what 3-month-olds do. And I'm a champion for babies at the end of the day and this is a clear win for babies. Babies are still slam dunking our most powerful computers when it comes to intuitive physics. Intagliata: Cognitive psychologist Susan Hespos of Northwestern University listed off a few examples of those "intuitive physics" principles.

Will I fail at this task? I grab a toddler food bowl and my coffee cup, holding them over my eyes. The face-scanning algorithm hunting for sunglasses is fooled. And I pocket a virtual $5.67, the fee I get for looking like a babysitter. Facial recognition algorithms are everywhere--in our iPhones, Instagram, Google Photos, and global network of surveillance cameras.

We live inside a body ruled by a brute, and the question for humanity may be whether we ever rise up and defy that brute. Such is, in rough outline, the key question of the human race's future in A Thousand Brains: A New Theory of Intelligence, the new book about artificial intelligence, and also, surprisingly, about human impulses, by Jeff Hawkins, which went on sale this week. "What's the purpose of living, why are we here, what would be a good goal for humanity," Hawkins mused during a conversation about the book with ZDNet via Zoom last week. "Intelligence is the thing that defines us, the thing we want to preserve and propagate." Life has evolved over millions of years to perpetuate genes via reproduction. Humans, like all life, are "unwitting servants" of genes, gifted with movement and ability only for the purpose of reproduction, writes Hawkins.

Machine Learning projects are known to fail frequently, according to Gartner 85% of all AI projects fail and even 96% fight with problems. Sure, when it comes to new technologies a high degree is normal, but these numbers are alarming. Typically, you read a lot about data quality, exaggerated expectations and wrong or non-existent goals. However, some of these issues can be avoided by assessing the projects in more detail before selecting a project for a Data Science/Machine Learning team. I would like to highlight some aspects from the perspective of an AI developer.

In 10 years, the circular economy will be the only economy, replacing wasteful linear economies, predicts Gartner. According to Gartner, circular economic business models encourage continuous reuse of materials to minimise waste and the demand for additional natural resource consumption. "The circular economy creates an ecosystem of materials," notes Sarah Watt, senior director analyst at Gartner. "What was previously viewed as waste now has value. However those ecosystems are complex, and include many interdependencies and feedback loops."

Artificial intelligence technologies can be used to help buildings and spaces track their waste in real-time and engage users by nudging them to correctly sort their waste. According to a study by the World Bank, 98% of the world's waste is sent to landfills, dumped into oceans or being incinerated, even though a high majority of daily consumables are recyclable. This is primarily due to the high level of contaminants found in recyclables, making previously clean material practically unrecyclable and financially unmarketable. In Toronto, for every percentage point decreased in contaminated waste can create up to $1 million in recycling cost savings every year, which can be attributed to the management and sorting costs incurred by the waste hauling and collection companies. Intuitive is a Canadian company which seeks to achieve zero waste through their AI solution, Oscar.

This week, Google announced that it has trained robots to sort through office trash and remove items that should go in recycling or the compost bin. The breakthrough came from Google X's Everyday Robot Project, an open-ended research initiative aimed at trying to integrate robots into daily life. Over the last few months, Google's office robots have decreased waste contamination levels from 20 percent to just five percent. Waste contamination levels are a measure of how much improperly sorted material there is mixed into the trash. For the trash sorting project, Google's decided that instead of trying to write complicated instructions for how to identify different kinds of items, they would see if the robots could just figure things out through trial and error.