Issue 73

Highlights

Could AI help fight conspiracy theories? – podcast

We’re used to hearing about the power of artificial intelligence to spread misinformation – but could it also be a tool for persuading people of the truth? Ian Sample speaks to Thomas Costello, an assistant professor of psychology at American University, who has published a study exploring the potential for AI chatbots to lead people away from conspiracy beliefs.

PyTorch Conference 2024 Recap: On Fire 🔥

The 2024 PyTorch Conference in San Francisco gathered nearly 1,500 AI researchers, developers, and enthusiasts. Over two days, the event featured engaging discussions, insightful keynotes, and hands-on sessions focused on artificial intelligence (AI) and advancements in PyTorch, the leading open-source machine learning framework.

Modeling relationships to solve complex problems efficiently

Associate Professor Julian Shun develops high-performance algorithms and frameworks for large-scale graph processing.

A co-lead on Sora, OpenAI’s video generator, has left for Google

One of the co-leads on OpenAI’s video generator, Sora, has left for Google. Tim Brooks, who was heading development on Sora with William Peebles, announced in a post on X that he’ll be joining Google DeepMind, Google’s AI research division, to work on video generation technologies and “world simulators.” “I had an amazing two years […]

New Responsible AI Features for Building Trustworthy AI

Discover the latest responsible AI features and capabilities to support building trustworthy AI solutions.

Paper of the week

Turning Word documents into PowerPoint slides

This is another fun but helpful machine-learning application: Turning Word documents into pleasing PowerPoint slides. This paper discusses combining embedding models, state machines, perceptrons, and sequence-to-sequence models to convert long Word documents into PowerPoint slides. It shows that combining multiple model architectures can yield some amazing results.

Video

OpenAI’s New ChatGPT In 3 Minutes! + NotebookLM and AlphaChip

Our paper on ray tracing: https://users.cg.tuwien.ac.at/zsolnai/gfx/adaptive_metropolis/ ❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambdalabs.com/papers

NVIDIA’s New AI: The King Is Here!

https://research.nvidia.com/labs/par/maskedmimic/ If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers

Articles

Data Formulator: Exploring how AI can help analysts create rich data visualizations

Data Formulator investigates combining UI interactions with natural language input. Powered by AI, it can help users create or adapt visualizations and supports continuous refinement through an iterative process. Now available on GitHub.

Scaling RAG from POC to Production

Whether you are building customer-facing chatbots to answer repetitive questions and reduce workload from customer service agents, or building a co-pilot for engineers to help them navigate complex user manuals step-by-step, RAG has become a key archetype of the application of LLMs. This has enabled LLMs to provide a contextually relevant response based on ground truth of hundreds or millions of documents, reducing hallucinations and improving the reliability of LLM-based applications.

Canvas is a new way to write and code with ChatGPT

Canvas was built with GPT-4o and can be manually selected in the model picker while in beta. Starting today we’re rolling out canvas to ChatGPT Plus and Team users globally. Enterprise and Edu users will get access next week. We also plan to make canvas available to all ChatGPT Free users when it’s out of beta.

The Rise of Pallas: Unlocking TPU Potential with Custom Kernels

Built as an extension to JAX and with dedicated support for PyTorch/XLA, Pallas enables the creation of custom kernels for GPU and TPU. In this post we move one step down the lineage of Greek mythology to Triton’s daughter, Pallas and discuss her namesake, the JAX extension for writing custom kernels for GPU and TPU.

Semantic Prompt Learning for Weakly-Supervised Semantic Segmentation

Weakly-Supervised Semantic Segmentation (WSSS) aims to train segmentation models using image data with only image-level supervision. Since precise pixel-level annotations are not accessible, existing methods typically focus on producing pseudo masks for training segmentation models by refining CAM-like heatmaps. However, the produced heatmaps may capture only the discriminative image regions of object categories or the associated co-occurring backgrounds.

Code

What's New In Python 3.13

It's Python 3.13 release day today. The big signature features are a better REPL with improved error messages, an option to run Python without the GIL and the beginnings of the new JIT. Here are some of the smaller highlights I spotted while perusing the release notes.