Cohere for AI's open-source multilingual large language model (LLM) Aya supports instructions in over 100 languages, addressing language representation gaps. Developed by 3000 researchers, this project used a high-quality dataset of prompt and completion pairs. The model, twice as linguistically encompassing as current models and proficient even in underrepresented languages, also captures cultural subtleties. Aya aims to boost language research, preserve lesser-known languages and cultures, and inspire collaborations. Model evaluation and adoption challenges persist.

Open-source AI model covering more than 100 languages released Axios

Mile Bluff Medical Center employs MEDITECH's AI-powered Expanse powered by Google Health to improve workflow, efficiency for over 150 users across care network. Incorporated into Mile Bluff's Expanse EHR, the tool utilizes natural language processing to extract info from structured/unstructured data, providing clinicians with timely, comprehensive view of patients. It also minimizes time spent on patient data cleaning, resonates across all departments, enables quick search for discrete data, and proves effective for chart preparation and abstracting. This radical collaboration highlights AI's potential in patient care efficiency.

Mile Bluff Medical Center now live with MEDITECH Expanse search and summarization powered by Google Health | MEDITECH Meditech

Ransomware attack on Romanian hospitals leads to over 100 facilities offline. Hackers demanded nearly $170,000 in Bitcoin. Romania's cybersecurity agency advised against payment and local company Bitdefender assists with recovery efforts.

More than 100 Romanian health care facilities taken offline after cyber attacks target hospitals NBC News

Healthcare leaders highlight the importance of safe AI use, especially as healthcare is set to be a major adopter of AI. They've built an AI platform to ensure safety and monitor its use. Potential dangers include increasing bias, data breaches, replacing human workers, and automating crucial tasks, but healthcare's cautious nature may restrict AI's risks. Automation of key judgments against the Hippocratic Oath may be avoided except for minor tasks like smart scheduling. Fear of bias and incomplete datasets causing harm exists, but health leaders' responsibility is to ensure AI's safe use within healthcare. AI can be viewed as another team member, always reviewed by a senior human.

'Bad science fiction'?: How health systems are keeping AI in check Beckers Hospital Review