AI - Artificial intelligence

cross-posted from: https://programming.dev/post/37407786

The ongoing development of artificial intelligence means that humans will simultaneously confront multiple interfaces of AI that exhibit a range of its propensity to contribute to good and bad, progress and destruction, and as act as a perpetrator of violence or a tool for peace. As a dual-use technology, AI can be adopted for military and civilian purposes alike.

Whether on the civilian or military side of adoption, AI contains inherent conflicts. Some of the main sources of conflict that policymakers must attempt to address are about how to ensure human rights values such as individual autonomy are preserved and not destroyed, navigate the organizational culture change necessary to respond to AI political end-uses, construct and upgrade the appropriate institutional arrangements needed for accountability, and ensure safeguards exist to enable trust-building.

Can demo here https://publicai.co/chat

In July, EPFL, ETH Zurich, and CSCS announced their joint initiative to build a large language model (LLM). Now, this model is available and serves as a building block for developers and organisations for future applications such as chatbots, translation systems, or educational tools.

The model is named Apertus – Latin for “open” – highlighting its distinctive feature: the entire development process, including its architecture, model weights, and training data and recipes, is openly accessible and fully documented.

AI researchers, professionals, and experienced enthusiasts can either access the model through the strategic partner Swisscom or download it from Hugging Face – a platform for AI models and applications – and deploy it for their own projects. Apertus is freely available in two sizes – featuring 8 billion and 70 billion parameters, the smaller model being more appropriate for individual usage. Both models are released under a permissive open-source license, allowing use in education and research as well as broad societal and commercial applications.

• SpikingBrain Technical Report: Spiking Brain-inspired Large Models.
• GitHub.

Humans navigate the social world by rapidly perceiving social features from other people and their interaction. Recently, large-language models (LLMs) have achieved high-level visual capabilities for detailed object and scene content recognition and description. This raises the question whether LLMs can infer complex social information from images and videos, and whether the high-dimensional structure of the feature annotations aligns with that of humans. We collected evaluations for 138 social features from GPT-4V for images (N = 468) and videos (N = 234) that are derived from social movie scenes. These evaluations were compared with human evaluations (N = 2,254). The comparisons established that GPT-4V can achieve human-like capabilities at annotating individual social features. The GPT-4V social feature annotations also express similar structural representation compared to the human social perceptual structure (i.e., similar correlation matrix over all social feature annotations). Finally, we modeled hemodynamic responses (N = 97) to viewing socioemotional movie clips with feature annotations by human observers and GPT-4V. These results demonstrated that GPT-4V based stimulus models can also reveal the social perceptual network in the human brain highly similar to the stimulus models based on human annotations. These human-like annotation capabilities of LLMs could have a wide range of real-life applications ranging from health care to business and would open exciting new avenues for psychological and neuroscientific research.

TL;DR.

LessWrong.

Research.

Toyota Research Institute Press Release.

Most American teens (72%) have had an experience with an AI chatbot, and over half use them several times a month. Character AI, one of the largest and most popular chatbot platforms, is available to children ages 13 and over. The platform hosts a wide variety of chatbots modeled after celebrities and fictional characters that appeal to children – both teens and younger kids. Several disturbing and tragic cases of extreme harm due to interactions on Character AI chatbots have already occurred since the company’s launch in September of 2022. As chatbots become more popular with children and teens, understanding the risks they present is critical to child safety online.

Adult researchers from ParentsTogether Action, in partnership with Heat Initiative, held 50 hours of conversation with Character AI chatbots using accounts registered to children. They found that Character AI chatbots engaged in a pattern of deeply concerning behaviors during these interactions. Harmful patterns and behaviors sometimes emerged within minutes of engagement.

Across 50 hours of conversation with 50 Character AI bots, ParentsTogether Action researchers logged 669 harmful interactions - an average of one harmful interaction every five minutes.

VibeVoice is a novel framework designed for generating expressive, long-form, multi-speaker conversational audio, such as podcasts, from text. It addresses significant challenges in traditional Text-to-Speech (TTS) systems, particularly in scalability, speaker consistency, and natural turn-taking. A core innovation of VibeVoice is its use of continuous speech tokenizers (Acoustic and Semantic) operating at an ultra-low frame rate of 7.5 Hz. These tokenizers efficiently preserve audio fidelity while significantly boosting computational efficiency for processing long sequences. VibeVoice employs a next-token diffusion framework, leveraging a Large Language Model (LLM) to understand textual context and dialogue flow, and a diffusion head to generate high-fidelity acoustic details. The model can synthesize speech up to 90 minutes long with up to 4 distinct speakers, surpassing the typical 1-2 speaker limits of many prior models.