The US Department of Homeland Security (DHS) has established an Artificial Intelligence Safety and Security Board, according to Secretary Alejandro Mayorkas. The board's inception is in response to President Biden's Executive Order for a “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.” The board's role is to create recommendations on how stakeholders of…
The Financial Times (FT) has granted OpenAI access to its news archives as part of a data sharing agreement, indicative of a growing trend among AI companies to secure private data sources. In this deal, OpenAI’s language model, ChatGPT, will provide summaries, direct quotes, and links to full FT articles, directing back to the original…
The Global AI Safety Summit held at the UK's Bletchley Park last year sparked considerable interest, with over 25 government representatives pledging to a joint declaration for collaborative AI oversight. However, expectations are realistic for the upcoming follow-up summit, scheduled for May 21-22, to be co-hosted virtually by the UK and South Korea. Significant players,…
French researchers have developed the first publicly available benchmark tool, 'DrBenchmark', to evaluate and standardize evaluation protocols for pre-trained masked language models (PLMs) in French, particularly in the biomedical field. Existing models lacked standardized protocols and comprehensive datasets, leading to inconsistent results and stalling progress in natural language processing (NLP) research.
The advent and advancement…
In the field of computational linguistics, large amounts of text data present a considerable challenge for language models, especially when specific details within large datasets need to be identified. Several models, like LLaMA, Yi, QWen, and Mistral, use advanced attention mechanisms to deal with long-context information. Techniques such as continuous pretraining and sparse upcycling help…
Emerging research from the New York University's Center for Data Science asserts that language models based on transformers play a key role in driving AI forward. Traditionally, these models have been used to interpret and generate human-like sequences of tokens, a fundamental mechanism used in their operational framework. Given their wide range of applications, from…
In collaboration with an artificial language network, neuroscientists at Massachusetts Institute of Technology (MIT) have revealed what type of sentences most significantly engage the brain’s primary language processing areas. The study indicates that sentences featuring unusual grammar or unexpected meaning trigger a heightened response in these language-oriented regions, as opposed to more straightforward phrases, which…
During the Festival of Learning 2024 at MIT, discussions were held on leveraging generative AI to enhance learning experiences for students both on and off campus. The panelists, comprising MIT faculty, instructors, staff, and students, emphasized that generative AI should be used to enrich, not replace, the educational experience. They highlighted the ongoing experimentation with…