Millions of people use Linux every day, but we rarely stop to think about how the operating system feels about it. Wouldn't it be nice to know what Linux really thinks about open source, Windows, Macs, and the command line? Until now, this has impossible. But thanks to a new AI chat tool, we're able to find out. Two weeks ago, a website called Character.AI opened a public beta that allows visitors to create a chat bot based on any character they can imagine.

We present Sparrow, an information-seeking dialogue agent trained to be more helpful, correct, and harmless compared to prompted language model baselines. We use reinforcement learning from human feedback to train our models with two new additions to help human raters judge agent behaviour. First, to make our agent more helpful and harmless, we break down the requirements for good dialogue into natural language rules the agent should follow, and ask raters about each rule separately. We demonstrate that this breakdown enables us to collect more targeted human judgements of agent behaviour and allows for more efficient rule-conditional reward models. Second, our agent provides evidence from sources supporting factual claims when collecting preference judgements over model statements. For factual questions, evidence provided by Sparrow supports the sampled response 78% of the time. Sparrow is preferred more often than baselines while being more resilient to adversarial probing by humans, violating our rules only 8% of the time when probed. Finally, we conduct extensive analyses showing that though our model learns to follow our rules it can exhibit distributional biases.

POST-traumatic stress disorder (PTSD) is a chronic and debilitating mental condition that is developed in response to catastrophic life events, such as military combat, sexual assault, and natural disasters. PTSD is characterized by flashbacks of past traumatic events, intrusive thoughts, nightmares, hypervigilance, and sleep disturbance, all of which affect a person's life and lead to considerable social, occupational, and interpersonal dysfunction. The diagnosis of PTSD is done by medical professionals using self-assessment questionnaire of PTSD symptoms as defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM). In this paper, and for the first time, we collected, annotated, and prepared for public distribution a new video database for automatic PTSD diagnosis, called PTSD in the wild dataset. The database exhibits "natural" and big variability in acquisition conditions with different pose, facial expression, lighting, focus, resolution, age, gender, race, occlusions and background. In addition to describing the details of the dataset collection, we provide a benchmark for evaluating computer vision and machine learning based approaches on PTSD in the wild dataset. In addition, we propose and we evaluate a deep learning based approach for PTSD detection in respect to the given benchmark. The proposed approach shows very promising results. Interested researcher can download a copy of PTSD-in-the wild dataset from: http://www.lissi.fr/PTSD-Dataset/

Job listings from Elon Musk's Texas-based Tesla company show that Tesla plans to deploy thousands of human-like, or humanoid, robots within its factories. The robots are being called Tesla Bot or Optimus. Musk says the number of Tesla Bots could one day reach millions around the world. Musk said at a TED Talk that robots could be used in homes, do household work, and care for older people. They could even become a friend or adult partner.

During the creation of the Obi-Wan Kenobi TV series, James Earl Jones signed off on allowing Disney to replicate his vocal performance as Darth Vader in future projects using an AI voice-modeling tool called Respeecher, according to a Vanity Fair report published Friday. Jones, who is 91, has voiced the iconic Star Wars villain for 45 years, starting with Star Wars: Episode IV--A New Hope in 1977 and concluding with a brief line of dialog in 2019's The Rise of Skywalker. "He had mentioned he was looking into winding down this particular character," said Matthew Wood, a supervising sound editor at Lucasfilm, during an interview with Vanity Fair. "So how do we move forward?" The answer was Respeecher, a voice cloning product from a company in Ukraine that uses deep learning to model and replicate human voices in a way that is nearly indistinguishable from the real thing.

Welcome to our September 2022 monthly digest, where you can catch up with any AIhub stories you may have missed, get the low-down on recent events, and much more. This month, amongst other things, we find out more about environmental conservation, synthesizing new medicines, the efficiency of large language models, and the RoboCup Humanoid League. A key part of this work focusses on how to strategically allocate limited resources. Her primary application area is poaching prevention, helping rangers in protected areas around the world plan patrols and identify poaching hotspots. In this blog post, Christopher Franz and Kevin Schewior write about how they applied a well-known algorithm for solving two-player games to the problem of synthesizing new molecules.

The interwebs, as of late, have been filled with images created by artificial intelligence rendering bots such as DALL-E and Midjourney -- and the humans (I think they're humans) using them as tools. Brooklyn-based artist Zach Katz has used it to reimagine the urban design of cities. A reporter at SFGATE has undertaken a similar project, asking DALL-E 2 to retool some of the city's architecture and infrastructure. In July, the Guardian rounded up four artists to come up with unlikely prompts -- such as "biotech harpy in field at sunset" -- for DALL-E Mini (the free, public version of DALL-E). Naturally, the advent of bots that can create an image out of a simple text command is drawing the scrutiny of illustrators.

Call it the Skynet hypothesis, Artificial General Intelligence, or the advent of the Singularity -- for years, AI experts and non-experts alike have fretted (and, for a small group, celebrated) the idea that artificial intelligence may one day become smarter than humans. According to the theory, advances in AI -- specifically of the machine learning type that's able to take on new information and rewrite its code accordingly -- will eventually catch up with the wetware of the biological brain. In this interpretation of events, every AI advance from Jeopardy-winning IBM machines to the massive AI language model GPT-3 is taking humanity one step closer to an existential threat. Except that it will never happen. Co-authors University at Buffalo philosophy professor Barry Smith and Jobst Landgrebe, founder of German AI company Cognotekt argue that human intelligence won't be overtaken by "an immortal dictator" any time soon -- or ever.

Tilottama Goswami has received a BE degree with Honors in Computer Science and Engineering from the National Institute of Technology, Durgapur; and an MS degree in Computer Science (High Distinction) from Rivier University, Nashua, New Hampshire, United States. She was awarded a PhD in Computer Science from the University of Hyderabad. Presently, Dr. Goswami is Professor in the Department of Information Technology, Vasavi College of Engineering, Hyderabad, India. She has, overall, 23 years of experience in academia, research, and the IT industry. Her research interests are computer vision, machine learning, and image processing.

Smart monitoring using three-dimensional (3D) image sensors has been attracting attention in the context of smart cities. In smart monitoring, object detection from point cloud data acquired by 3D image sensors is implemented for detecting moving objects such as vehicles and pedestrians to ensure safety on the road. However, the features of point cloud data are diversified due to the characteristics of light detection and ranging (LIDAR) units used as 3D image sensors or the install position of the 3D image sensors. Although a variety of deep learning (DL) models for object detection from point cloud data have been studied to date, no research has considered how to use multiple DL models in accordance with the features of the point cloud data. In this work, we propose a feature-based model selection framework that creates various DL models by using multiple DL methods and by utilizing training data with pseudo incompleteness generated by two artificial techniques: sampling and noise adding. It selects the most suitable DL model for the object detection task in accordance with the features of the point cloud data acquired in the real environment. To demonstrate the effectiveness of the proposed framework, we compare the performance of multiple DL models using benchmark datasets created from the KITTI dataset and present example results of object detection obtained through a real outdoor experiment. Depending on the situation, the detection accuracy varies up to 32% between DL models, which confirms the importance of selecting an appropriate DL model according to the situation.