Global AI Weekly
Issue number: 80 | Tuesday, January 14, 2025
Highlights
}
Microsoft forms new internal dev-focused AI org
Microsoft has created a new engineering organization aimed at accelerating AI infrastructure and software development within the company. According to Bloomberg, Jay Parikh, previously VP and global head of engineering at Meta, will lead the new division.
techcrunch.com
}
Elon Musk says all human data for AI training ‘exhausted’
Tech boss suggests move to self-learning synthetic data though some warn this could cause ‘model collapse’ Business live – latest updatesArtificial intelligence companies have run out of data for training their models and have “exhausted” the sum of human knowledge, Elon Musk has said.
theguardian.com
}
OpenAI presents its preferred version of AI regulation in a new ‘blueprint’
OpenAI on Monday published what it’s calling an “economic blueprint” for AI: a living document that lays out policies the company thinks it can build on with the U.S. government and its allies. The blueprint, which includes a forward from Chris Lehane, OpenAI’s VP of global affairs, asserts that the U.S. must act to attract […]
techcrunch.comResearch
}
DRC-Coder: Automated DRC Checker Code Generation Using LLM Autonomous Agent
In the advanced technology nodes, the integrated design rule checker (DRC) is often utilized in place and route tools for fast optimization loops for power-performance-area.
research.nvidia.comVideo
}
Marco Casalaina - Silicon Minds, Human Hearts
Join us at the Garage in Microsoft San Francisco as Marco shares his insights on the transformative impact of AI in the enterprise world.
youtube.com
}
Generative AI Level Up Tuesdays
From Azure AI Foundry and Azure AI Agents to exploring responsible AI tooling and beyond, there's something for every developer. Don't miss out on this opportunity to join us and enhance your AI skills!
developer.microsoft.comArticles
}
How to Build a Knowledge Graph in Minutes (And Make It Enterprise-Ready)
I tried and failed creating one—but it was when LLMs were not a thing!
towardsdatascience.comDeliberative alignment: reasoning enables safer language models
Deliberative alignment: reasoning enables safer language models Introducing our new alignment strategy for o1 models, which are directly taught safety specifications and how to reason over them.
openai.com
}
Teaching AI to communicate sounds like humans do
Inspired by the mechanics of the human vocal tract, a new AI model can produce and understand vocal imitations of everyday sounds. The method could help build new sonic interfaces for entertainment and education.
news.mit.edu
}
Understanding the Evolution of ChatGPT: Part 2 — GPT-2 and GPT-3
Before diving into the details of GPT-2 and GPT-3, let’s first take a look at the three key elements that have influenced their model design: task-agnostic learning, the scale hypothesis, and in-context learning.
towardsdatascience.com
}
Ethical Implications of AI in Society
to permeate various aspects of daily life, it raises critical questions about fairness, privacy, autonomy, and accountability.
medium.com
}
LlamaIndex Typescript now supports Azure AI Search as a vector store
Learn how JavaScript and TypeScript developers can integrate LlamaIndex with Azure AI Search to build smarter apps that find and use relevant data for better...
techcommunity.microsoft.com
}
Building Deterministic GenAI Chatbots In Regulated Industries
How to Build Reliable and Contextual AI Chatbots for Finance, Healthcare and Other Industries.
towardsdatascience.com
}
Love AI or Hate It? AI IS Reshaping the World in 2025
Artificial Intelligence (AI) is no longer a buzzword — it’s essential to how businesses, creators, and entrepreneurs operate today. Tools…
medium.comUpcoming Events
}
Global AI Bootcamp 2025
The Global AI Bootcamp is an annual event that occurs worldwide, where developers and AI enthusiasts can learn about AI through workshops, sessions, and discussions. Local chapters of the Global AI…
globalai.communityCode
Codestral 25.01
Brand new code-focused model from Mistral. Unlike the first Codestral this one isn't (yet) available as open weights. The model has a 256k token context - a new record for Mistral.
simonwillison.net
}
Integrating Ascend Backend with Torchtune through PyTorch Multi-Device Support
In this blog, we will briefly introduce torchtune, the Ascend backend, and demonstrate how torchtune can be used to fine-tune models with Ascend.
pytorch.org