Large language models (LLMs) are machine-learning models designed to understand and generate human language. State-of-the-art LLMs have demonstrated outstanding potential in open-domain question answering (ODQA), where the model is tasked with providing answers to factual questions.Large language models (LLMs) are machine-learning models designed to understand and generate human language. State-of-the-art LLMs have demonstrated outstanding potential in open-domain question answering (ODQA), where the model is tasked with providing answers to factual questions.[#item_full_content]

Researchers at Apple Computer Company have found evidence, via testing, showing that the seemingly intelligent responses given by AI-based LLMs are little more than an illusion. In their paper posted on the arXiv preprint server, the researchers argue that after testing several LLMs, they found that they are not capable of performing genuine logical reasoning.Researchers at Apple Computer Company have found evidence, via testing, showing that the seemingly intelligent responses given by AI-based LLMs are little more than an illusion. In their paper posted on the arXiv preprint server, the researchers argue that after testing several LLMs, they found that they are not capable of performing genuine logical reasoning.[#item_full_content]

Penn Engineers have developed a new algorithm that allows robots to react to complex physical contact in real time, making it possible for autonomous robots to succeed at previously impossible tasks, like controlling the motion of a sliding object.Penn Engineers have developed a new algorithm that allows robots to react to complex physical contact in real time, making it possible for autonomous robots to succeed at previously impossible tasks, like controlling the motion of a sliding object.[#item_full_content]

While artificial intelligence (AI) bots can serve a legitimate purpose on social media—such as marketing or customer service—some are designed to manipulate public discussion, incite hate speech, spread misinformation or enact fraud and scams. To combat potentially harmful bot activity, some platforms have published policies on using bots and created technical mechanisms to enforce those policies.While artificial intelligence (AI) bots can serve a legitimate purpose on social media—such as marketing or customer service—some are designed to manipulate public discussion, incite hate speech, spread misinformation or enact fraud and scams. To combat potentially harmful bot activity, some platforms have published policies on using bots and created technical mechanisms to enforce those policies.[#item_full_content]

Sometimes, we try to capture a QR code with a good digital camera on a smartphone, but the reading eventually fails. This usually happens when the QR code itself is of poor image quality, or if it has been printed on surfaces that are not flat—deformed or with irregularities of unknown pattern—such as the wrapping of a courier package or a tray of prepared food.Sometimes, we try to capture a QR code with a good digital camera on a smartphone, but the reading eventually fails. This usually happens when the QR code itself is of poor image quality, or if it has been printed on surfaces that are not flat—deformed or with irregularities of unknown pattern—such as the wrapping of a courier package or a tray of prepared food.[#item_full_content]

A world of color and texture could soon become more accessible to people who are blind or have low vision, via new software that narrates what a camera records.A world of color and texture could soon become more accessible to people who are blind or have low vision, via new software that narrates what a camera records.[#item_full_content]

From the water that comes out of the faucet to the chemical reactions in jet engines that propel planes, turbulence affects our everyday lives. Researchers at Georgia Tech are studying the complex physics of turbulence in simplified settings that could help us better understand nature and engineering.From the water that comes out of the faucet to the chemical reactions in jet engines that propel planes, turbulence affects our everyday lives. Researchers at Georgia Tech are studying the complex physics of turbulence in simplified settings that could help us better understand nature and engineering.[#item_full_content]

A team of engineers at Apple has developed an AI-based model called Depth Pro that can map the depth of a 2D image. The team has written a paper describing the app and its capabilities and has posted it on the arXiv preprint server. They have also posted an announcement regarding the app on the company’s Machine Learning Research page.A team of engineers at Apple has developed an AI-based model called Depth Pro that can map the depth of a 2D image. The team has written a paper describing the app and its capabilities and has posted it on the arXiv preprint server. They have also posted an announcement regarding the app on the company’s Machine Learning Research page.[#item_full_content]

Artificial Intelligence has learned to master language, generate art, and even beat grandmasters at chess. But can it crack the code of abstract reasoning—those tricky visual puzzles that leave humans scratching their heads?Artificial Intelligence has learned to master language, generate art, and even beat grandmasters at chess. But can it crack the code of abstract reasoning—those tricky visual puzzles that leave humans scratching their heads?[#item_full_content]

Neural networks have a remarkable ability to learn specific tasks, such as identifying handwritten digits. However, these models often experience “catastrophic forgetting” when taught additional tasks: They can successfully learn the new assignments, but “forget” how to complete the original. For many artificial neural networks, like those that guide self-driving cars, learning additional tasks thus requires being fully reprogrammed.Neural networks have a remarkable ability to learn specific tasks, such as identifying handwritten digits. However, these models often experience “catastrophic forgetting” when taught additional tasks: They can successfully learn the new assignments, but “forget” how to complete the original. For many artificial neural networks, like those that guide self-driving cars, learning additional tasks thus requires being fully reprogrammed.[#item_full_content]