AI Innovations: Berkeley’s Sky-T1 Model and Microsoft’s New AI Division

Explore UC Berkeley's groundbreaking Sky-T1-32B model, an open-source AI trained for under $450 that challenges industry giants. Discover Microsoft's strategic move to form a new AI-focused division, CoreAI, to accelerate innovation. Plus, delve into OpenAI's regulatory blueprint and Midjourney's decision to block images of political figures, highlighting the ethical challenges in AI development.

Sources:
https://www.marktechpost.com/2025/01/13/uc-berkeley-researchers-released-sky-t1-32b-preview-an-open-source-reasoning-llm-trained-for-under-450-surpasses-openai-o1-on-benchmarks-like-math500-aime-and-livebench/
https://techcrunch.com/2025/01/13/microsoft-forms-new-internal-dev-focused-ai-org/
https://techcrunch.com/2025/01/13/openai-presents-its-preferred-version-of-ai-regulation-in-a-new-blueprint/
https://apnews.com/article/midjourney-ai-imagegenerator-biden-trump-deepfakes-bc6c254ddb20e36c5e750b4570889ce1

Outline:
(00:00:00) Introduction
(00:00:42) UC Berkeley Researchers Released Sky-T1-32B-Preview: An Open-Source Reasoning LLM Trained for Under $450 Surpasses OpenAI-o1 on Benchmarks like Math500, AIME, and Livebench
(00:03:52) Microsoft forms new internal dev-focused AI org
(00:07:13) OpenAI presents its preferred version of AI regulation in a new ‘blueprint’
(00:10:41) AI image-generator Midjourney blocks images of Biden, Trump