Issue 25

Highlights

OpenAI forms team to study ‘catastrophic’ AI risks, including nuclear threats

OpenAI today announced that it’s created a new team to assess, evaluate and probe AI models to protect against what it describes as “catastrophic risks.” The team, called Preparedness, will be led by Aleksander Madry, the director of MIT’s Center for Deployable Machine Learning. (Madry joined OpenAI in May as “head of Preparedness,” according to […] © 2023 TechCrunch. All rights reserved. For personal use only.

Now add a walrus: Prompt engineering in DALL-E 3

Last year I wrote about my initial experiments with DALL-E 2, OpenAI's image generation model. I've been having an absurd amount of fun playing with its sequel, DALL-E 3 recently. Here are some notes, including a peek under the hood and some notes on the leaked system prompt.

Making genetic prediction models more inclusive

MIT computer scientists developed a way to calculate polygenic scores that makes them more accurate for people across diverse ancestries.

Video

Content Safety - Global AI Notes

Join us as we talk with Sarah Bird - Azure AI Content Safety is a new service that helps you detect and filter harmful user-generated and AI-generated content in your applications and services. It can monitor both text and image context across categories and languages and provides severity scores to help prioritize content for review. Azure AI Content Safety can be leveraged through our API, or interactive studio, as well as through built in features across products and services such as Azure OpenAI, Copilot and Bing.

Maarten Goet - Humans In AI

In this video, Maarten Goet, a Microsoft MVP, RD from the Netherlands, and CTO at Wortell, shares his insights on the revolutionary potential of AI. He compares its transformative capacity to the impact of UI in the 90s and the internet in the 2000s.

Articles

Multi-modal prompt injection image attacks against GPT-4V

GPT4-V is the new mode of GPT-4 that allows you to upload images as part of your conversations. It's absolutely brilliant. It also provides a whole new set of vectors for prompt injection attacks.

The NVIDIA & Microsoft Partnership: A New Era in Computing

Introduction Hold onto your hats, folks! The tech world is buzzing with news that could reshape the computing landscape as we know it. NVIDIA, a powerhouse in AI computing chips, is teaming up with Microsoft to challenge Intel’s long-standing dominance in the personal computer market. So, what does this partnership mean for the industry, consumers, and the future of computing? Let’s dive in.

Understanding OpenAI’s Preparedness Team: A Step Towards Safer AI Development

In the rapidly evolving world of artificial intelligence (AI), staying ahead of potential risks and ensuring safe development is paramount. OpenAI, a leading entity in the AI space, has recently introduced its Preparedness Team, a strategic move to assess and mitigate the risks associated with highly advanced foundation models.

Humanity at risk from AI ‘race to the bottom’, says tech expert

MIT professor behind influential letter says unchecked development is allowing a few AI firms to jeopardise society’s future.

Compiling NumPy code into C++ or CUDA via torch.compile

Quansight engineers have implemented support for tracing through NumPy code via torch.compile in PyTorch 2.1. This feature leverages PyTorch’s compiler to generate efficient fused vectorized code without having to modify your original NumPy code. Even more, it also allows for executing NumPy code on CUDA just by running it through torch.compile under torch.device("cuda")!

Upcoming events

12 December - Global AI Conference 23

The Global AI Conference is an event that brings together AI fans from all over the world. It's put together by the Global AI Community. You can jump into the worldwide livestream, check out a local conference in person, or chat and connect with experts on our Discord Channel. It's a great opportunity to dive deep into the world of AI, so don't miss out!

Register now for Make Azure AI Real YouTube technical series

Microsoft Reactor’s Make Azure AI Real series is a collection of AI livestreams that are focused on AI technical content from experts at Microsoft. You will learn how to use Microsoft AI tools and services, such as Azure Cognitive Services, Azure OpenAI Service, and more. You will also get insights into the latest AI trends and best practices from Microsoft AI experts. Join us for this exciting series and discover how to make AI real for your projects and goals.

Register Here: Event Series | Microsoft Reactor