Researchers have discovered that transformer models develop 'privileged bases' - specific coordinate axes in their residual streams. These axes are ti...
Researchers have proposed a new method called Dual-Stream Cross-Anchor Correction (DSCC) to improve the performance of multimodal large language model...
ChatGPT Ads, a platform for advertising on the AI chatbot, is expanding its reach to 31 European countries. This move allows advertisers to target use...
OpenAI has launched an initiative aimed at improving democratic oversight of artificial intelligence (AI) in national security. The program provides s...
A team of researchers from IBM's T.J. Watson Research Center has proposed a new method for determining the optimal memory usage for artificial agents....
OpenAI is teaming up with CodeAI to educate students on how to develop and use artificial intelligence responsibly. The partnership aims to increase A...
OpenAI is implementing new safeguards to monitor and align its frontier AI models with safety and security standards. The company aims to balance inno...
OpenAI has introduced a version of its ChatGPT chatbot specifically designed for teenagers. This version is intended to help teens learn and think cri...
Asana, a project management software company, claims to have replaced an outdated testing system with the help of OpenAI's Codex AI model. According t...
Researchers have replicated an earlier study on AI efficiency assessment using Floating Point Operations (FLOPs). They found that while FLOPs can be u...
Researchers developed a controlled test to evaluate the performance of large language models (LLMs) in medical reasoning. They created synthetic patie...
Researchers have introduced a new challenge called The Unwritten Benchmark to test multimodal machine learning models' ability to perform abstract per...