Articles for category: AI News

March 15, 2025

ikayaniaamirshahzad@gmail.com

Last Week in AI #300

Editorial Note: Hi, Andrey here (the person running this newsletter and co-hosting the podcast). I’d like to apologize for not having been releasing the text newsletter so far in 2025, especially to those of you who don’t listen to the podcast but like reading this, and who support this Substack financially. I wanted to do something special for #300, but repeatedly could not find the time and put it off. At long last, I decided to just put out this normal edition to get things back on track; starting now, you can expect to get the text newsletter once a

March 15, 2025

ikayaniaamirshahzad@gmail.com

Getting Government AI Engineers to Tune into AI Ethics Seen as Challenge 

By John P. Desmond, AI Trends Editor   Engineers tend to see things in unambiguous terms, which some may call Black and White terms, such as a choice between right or wrong and good and bad. The consideration of ethics in AI is highly nuanced, with vast gray areas, making it  challenging for AI software engineers to apply it in their work.   That was a takeaway from a session on the Future of Standards and Ethical AI at the AI World Government conference held in-person and virtually in Alexandria, Va. this week.    An overall impression from the conference is that the discussion of AI

March 15, 2025

ikayaniaamirshahzad@gmail.com

Chinese AI jewel Deepseek reportedly restricts employee travel amid AI security concerns

Summary Deepseek employees working on AI models must surrender their passports and can no longer travel freely abroad, according to insiders. Whether these restrictions come from the company or Chinese authorities remains unclear. In Zhejiang province, where Deepseek’s parent company is based, government officials are now screening potential investors before allowing meetings with company management. These measures appear aimed at preventing data leaks and unauthorized acquisitions. These measures contrast sharply with Deepseek’s public image as an open-source advocate and underdog promoting free access to AI models. The company’s visibility has increased significantly since its R1 breakthrough. CEO Liang Wenfeng now

March 15, 2025

ikayaniaamirshahzad@gmail.com

NVIDIA’s Hybrid: Combining Attention and State Space Models for Breakthrough Performance of Small Language Models

Language models (LMs) based on transformers have become the gold standard in natural language processing, thanks to their exceptional performance, parallel processing capabilities, and ability to retain long-term context via key-value (KV) caches. However, these benefits come at a cost—transformers require quadratic computational resources and large memory footprints, presenting significant efficiency challenges. On the other hand, state space models (SSMs), such as Mamba, boast constant computational complexity and hardware-friendly design, but they struggle with memory recall, which hampers their performance on diverse language tasks. To address the abovementioned issues, in a new paper Hymba: A Hybrid-head Architecture for Small Language

March 15, 2025

ikayaniaamirshahzad@gmail.com

A new programming language for high-performance computing, with much less code

Paper overview Credit: arXiv (2024). DOI: 10.48550/arxiv.2411.07211 Many companies invest heavily in hiring talent to create the high-performance library code that underpins modern artificial intelligence systems. NVIDIA, for instance, developed some of the most advanced high-performance computing (HPC) libraries, creating a competitive moat that has proven difficult for others to breach. But what if a couple of students, within a few months, could compete with state-of-the-art HPC libraries with a few hundred lines of code, instead of tens or hundreds of thousands? That’s what researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have shown with a new programming

March 15, 2025

ikayaniaamirshahzad@gmail.com

Interview with Tunazzina Islam: Understand microtargeting and activity patterns on social media

In this interview series, we’re meeting some of the AAAI/SIGAI Doctoral Consortium participants to find out more about their research. The Doctoral Consortium provides an opportunity for a group of PhD students to discuss and explore their research interests and career objectives in an interdisciplinary workshop together with a panel of established researchers. In the third of our interviews with the 2025 cohort, we heard from Tunazzina Islam who has recently completed her PhD in Computer Science at Purdue University, advised by Dr Dan Goldwasser. Her primary research interests lie in computational social science (CSS), natural language processing (NLP), and

March 15, 2025

ikayaniaamirshahzad@gmail.com

LaTeXify in Python: No Need to Write LaTeX Equations Manually

In mathematical computing and scientific programming, clear and precise representation of functions is essential. While LaTeX is widely used for formatting mathematical expressions, manually writing equations can be time-consuming. The latexify-py library offers a solution by automatically converting Python functions into LaTeX-formatted expressions. This functionality enhances both readability and documentation by providing a structured and visually coherent representation of mathematical operations. Setting the Stage: Installing latexify-py Before we embark on this fusion of code and creative math, let’s install the package: !pip install latexify-py This package allows us to convert Python functions into LaTeX-rendered equations with minimal effort. Once installed,

March 15, 2025

ikayaniaamirshahzad@gmail.com

Patronus AI Introduces the Industry’s First Multimodal LLM-as-a-Judge (MLLM-as-a-Judge): Designed to Evaluate and Optimize AI Systems that Convert Image Inputs into Text Outputs

​In recent years, the integration of image generation technologies into various platforms has opened new avenues for enhancing user experiences. However, as these multimodal AI systems—capable of processing and generating multiple data forms like text and images—expand, challenges such as “caption hallucination” have emerged. This phenomenon occurs when AI-generated descriptions of images contain inaccuracies or irrelevant details, potentially diminishing user trust and engagement. Traditional methods of evaluating these systems often rely on manual inspection, which is neither scalable nor efficient, highlighting the need for automated and reliable evaluation tools tailored to multimodal AI applications.​ Addressing these challenges, Patronus AI has

March 15, 2025

ikayaniaamirshahzad@gmail.com

Clickhouse Acquires HyperDX To Advance Open-Source Observability

ClickHouse, known for its high-speed analytical database, has acquired HyperDX, an open-source observability platform built on its technology. The acquisition integrates HyperDX’s UI and session replay capabilities with ClickHouse’s database performance.  A core competence of ClickHouse is its ability to analyze large-scale datasets in real time, making it a popular choice for users that require fast data insights and have demanding analytics workloads.  The strategic acquisition of HyperDX enables ClickHouse to expand its role in open-source observability and address previous gaps in its ecosystem, such as the lack of user-friendly interfaces and out-of-the-box observability features offered by more established solutions.