News

Research scientists from Canada's National Research Council have extensively tested four large vision-language models (LVLM) for potential bias in gender and race. Despite alignment efforts to block unwanted or harmful responses, it is challenging, if not impossible, to create a completely impartial AI model. This bias comes from training these AI models on copious amounts…

In the field of natural language processing (NLP), large language models (LLMs) have revolutionized how machines understand and generate human-like text. Their application, however, is often limited by their hefty demand for computational resources. This problem has led researchers to test smaller, more compact LLMs, particularly their abilities to efficiently summarize meeting transcripts. Historically, text and…

Google's CEO Sundar Pichai is championing a shift from traditional search engines to more interactive and multi-faceted tools for information retrieval, as evident in his interview with WIRED. Pichai has given an insight into Google's new project, Gemini, an AI chatbot developed not only as an alternative to search engines but also as competition to…

This week in artificial intelligence (AI) news, debate continued around weaponizing AI, regulating the technology, battling fake images, and the environmental impact of AI power usage. A recent study revealed that AI models in war simulations are often quick to resort to nuclear weapons. AI is being increasingly utilized in defense sectors, from weaponry to…

The continued evolution of computational science has given rise to physics-informed neural networks (PINNs), a cutting-edge method for solving forward and inverse problems governed by partial differential equations (PDEs). PINNs uniquely incorporate physical laws into the learning process, leading to a substantial increase in predictive accuracy and robustness. However, as PINNs become more in-depth and…

Large Language Models (LLMs) have become critical tools for Natural Language Processing (NLP) tasks, including question-answering, text summarization, and few-shot learning. Despite their prevalence, the development process of the more potent models, particularly their pretraining data composition, often remains undisclosed. This tendency towards opacity complicates our understanding of how the pretraining corpus influences a model's…

Retrieval-augmented language models often only obtain small sections from a corpus, inhibiting their potential to adapt to global changes and incorporate extensive knowledge. This problem is prevalent in most existing methods that struggle to leverage large-scale discourse structure effectively. It is notably significant for thematic questions that require knowledge integration from multiple text sections. Large Language…

At the Frontiers of General Artificial Intelligence Technology Exhibition in Beijing, a virtual robot toddler called Tong Tong was revealed. Developed by the Beijing Institute for General Artificial Intelligence (BIGAI), Tong Tong is not a physical robot, but a virtual one. Visitors at the exhibition saw how Tong Tong interacted and adapted to her environment…

The 2024 Post-Industrial Summit, scheduled for February 28-29 in Menlo Park, California, will bring business leaders together to discuss the role of AI in reshaping the future of businesses. The summit is being hosted by the Post-Industrial Institute and SRI International and will include insights from experts from AWS, SAP, Salesforce, SRI, Broadcom, Swisscom, Deloitte,…

Speech recognition technology has become essential in various applications, helping machines to recognize and process human speech. Achieving accurate recognition across different languages and dialects is challenging due to factors like accents, intonation, and background noise. Various methods have been tried to enhance speech recognition systems, including the use of complex architectures like Transformers, which…

Language models are the cornerstone of many applications and breakthroughs in artificial intelligence, driving progress in machine translation, content creation, and conversational AI. However, the scale and size of these models often impose significant computational demands, raising concerns about accessibility and environmental impact due to high energy consumption and carbon emissions. A key element of improving…