engblogs

Custom LLMs can enhance operations and protect IP by using organizations' own data for AI models, with a focus on privacy and compliance.

Introducing Turbo v2, our fastest model yet, offering speech generation at ≈400ms latency and mulaw 8khz output support.

LLM distillation techniques will gain prominence in 2024 as data science teams prioritize the use of smaller, deployable models to enhance performance and cost-efficiency.

Responsible AI at Google Research: Context in AI Research (CAIR) explores the development of AI methods with a focus on responsibility, fairness, transparency, and inclusion.

Overcoming leakage on error-corrected quantum processors: This blogpost discusses the challenges of leakage in quantum error correction and introduces a new operation, data qubit leakage removal (DQLR), that effectively mitigates leakage and improves the performance of quantum error correction.

Introducing OpenAI Data Partnerships to produce public and private datasets for AI model training to enhance understanding of various domains and improve AI's usefulness.

LLM distillation techniques will play a crucial role in data science teams' practices in 2024, enabling the creation of compact, production-ready models and accelerating data labeling.

Discover 3 new ways Duet AI can enhance your productivity in the Google Cloud console.

Google supports launch of DAOS Foundation for next-generation HPC and AI storage

Introducing Cloud SQL in-place upgrade: move from Enterprise to Enterprise Plus with ease

Improve autoscaling for your streaming jobs with Dataflow Streaming Engine autotuning

Google Cloud discusses their approach to trust and transparency in AI, focusing on responsible development, security measures, privacy safeguards, and compliance efforts.