Issue 68

Highlights

‘He was in mystic delirium’: was this hermit mathematician a forgotten genius whose ideas could transform AI – or a lonely madman?

In isolation, Alexander Grothendieck seemed to have lost touch with reality, but some say his metaphysical theories could contain wonders One day in September 2014, in a hamlet in the French Pyrenean foothills, Jean-Claude, a landscape gardener in his late 50s, was surprised to see his neighbour at the gate. He hadn’t spoken to the 86-year-old in nearly 15 years after a dispute over a climbing rose that Jean-Claude had wanted to prune. The old man lived in total seclusion, tending to his garden in the djellaba he always wore, writing by night, heeding no one. Now, the long-bearded seeker looked troubled.

Apple and Nvidia could be OpenAI’s next big investors

Nvidia and Apple are reportedly in talks to contribute to OpenAI’s next fundraising round — a round that could value the ChatGPT maker at $100 billion. Per its sources, the New York Times says earlier OpenAI investor Thrive Capital would lead the deal should it happen. (Bloomberg was the first to report Nvidia’s involvement.

Paper of the week

Estimating demand for deliveries in a city

How do you plan drivers, cars, and package deliveries in a busy city? It's a hard problem to solve. Efforts are being made to try and solve part of the equation with AI. Estimating demand across a city with a graph-based approach. A fun to read paper with some pretty cool ideas.

Video

Black Box: episode 6 – Shut it down? - podcast

Revisited: Guardian journalist Michael Safi delves into the world of artificial intelligence, exploring the dangers and promises it holds for society This week we are revisiting the Black Box series. This episode was first broadcast on 21 March 2024.

Crazy New AI Learned To Rewrite Doom!

My paper on simulations that look almost like reality is available for free here: If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers

A Helping Hand for LLMs (Retrieval Augmented Generation) - Computerphile

Mike is based at the University of Nottingham's School of Computer Science.

Articles

Build, tweak, repeat

Making it easier to develop and share generative AI applications.

Collaborators: AI and the economy with Brendan Lucier and Mert Demirer

Researcher Brendan Lucier and professor Mert Demirer are applying their micro- and macroeconomic expertise, respectively, to forecasting the economic impact of AI. They share how they’re using a task-level breakdown of occupations to help predict the future.

An Easy Way to Remove Tourists from Photos

Image cleanup with Python, PIL, and OpenCV

How to Evaluate Jailbreak Methods: A Case Study with the StrongREJECT Benchmark

When we began studying jailbreak evaluations, we found a fascinating paper claiming that you could jailbreak frontier LLMs simply by translating forbidden prompts into obscure languages. Excited by this result, we attempted to reproduce it and found something unexpected.

Anatomy of a Textual User Interface

Will McGugan used Textual and my LLM Python library to build a delightful TUI for talking to a simulation of Mother, the AI from the Aliens movies.

Accelerate Your AI: PyTorch 2.4 Now Supports Intel GPUs for Faster Workloads

We have exciting news! PyTorch 2.4 now supports Intel® Data Center GPU Max Series and the SYCL software stack, making it easier to speed up your AI workflows for both training and inference. This update allows for you to have a consistent programming experience with minimal coding effort and extends PyTorch’s device and runtime capabilities, including device, stream, event, generator, allocator, and guard, to seamlessly support streaming devices. This enhancement simplifies deploying PyTorch on ubiquitous hardware, making it easier for you to integrate different hardware back ends.

A framework for solving parabolic partial differential equations

A new algorithm solves complicated partial differential equations by breaking them down into simpler problems, potentially guiding computer graphics and geometry processing.

Code

How to Secure Azure AI Models Using Managed Identity

In the world of cloud computing, securing your AI models is crucial. Azure offers powerful features to protect your models from…

Training AI Models on CPU

To simplify matters, PyTorch offers the torch.backends.xeon.run_cpu script for automatically configuring the memory and threading libraries so as to optimize runtime performance. Importantly, the impact of the optimizations that we discuss on runtime performance is likely to vary greatly based on the model and the details of the environment (e.g., see the high degree of variance between models on the official PyTorch TouchInductor CPU Inference Performance Dashboard ).