Good morning! For developers today, GitHub is synonymous with hosting source code repositories, but it wasn't always the go-to platform. DeepMind has developed SAFE, an AI system designed to automatically fact-check outputs from large language models like ChatGPT. Wind River has launched its Studio Developer platform on Amazon Web Services, allowing embedded developers to leverage the cloud's scalability and collaboration capabilities.

— Forrest Knight & Meghanadh Vasireddy

GitHub Take Over as the Go-To for Code Sharing

For developers today, GitHub is synonymous with hosting source code repositories. However, it wasn't always the go-to platform. Before GitHub's launch in 2008, open-source projects were primarily hosted on SourceForge and Google Code—sites focused more on software distribution than collaborative development.

GitHub's innovative approach introduced:

• User profiles

• Code browsing

• Issue tracking

• Pull requests for contributions

This fostered a thriving community for sharing and contributing to projects. Crucially, GitHub also provided native Git support from the start, aligning with the open-source world's shift towards the distributed version control system created by Linus Torvalds in 2005.

Rapid Growth: Within just a few years, GitHub's effortless Git hosting and social coding features led to explosive growth. The Ruby on Rails community was an early advocate, accelerating GitHub's adoption. By 2011, GitHub had surpassed SourceForge and Google Code in both repositories and users.

While major tech giants like Google and Facebook built internal tools, GitHub became the de facto standard for most companies due to:

• Its specialization in Git hosting

• Open-source collaboration features

Read More Here

SAFE: DeepMind's AI Fact-Checker for LLMs

Large language models (LLMs) like ChatGPT are amazing at generating human-like text, answering questions, and solving problems. However, they struggle with accuracy, requiring manual fact-checking which limits their usefulness.

DeepMind has developed SAFE (Search-Augmented Factuality Evaluator), an AI system designed to automatically fact-check LLM outputs. Here's how it works:

• SAFE uses an LLM component to break down a response into individual facts/claims

• It sends queries to Google Search and cross-references the results to evaluate each fact's accuracy

• This approach mimics how humans fact-check using search engines

In rigorous testing on approximately 16,000 facts from various LLMs, SAFE demonstrated impressive performance:

• SAFE's factuality judgments aligned with human evaluators 72% of the time

• When disagreeing with humans, SAFE was more accurate 76% of the time

Read More Here

Wind River Brings Studio Developer to AWS Cloud

Wind River has launched its Studio Developer platform on Amazon Web Services (AWS), allowing embedded developers to leverage the scalability and collaboration capabilities of the cloud.

Studio Developer is Wind River's DevOps solution for building safety-critical intelligent edge systems. By hosting it on AWS, customers can take advantage of cloud-scale automated build and testing processes tailored for embedded development.

The deployment showcased at the AWS Santa Clara lab demonstrates cloud-native over-the-air updates for software-defined vehicles. It utilizes AWS services like Amazon EC2 for the build pipeline and incorporates generative AI through integration with:

• Amazon Q virtual assistant

• Amazon CodeWhisperer code generation tool

Key benefits of the AWS-hosted Studio Developer include:

• Scalable cloud resources for compute-intensive tasks like builds and tests

• Collaborative development environment accessible from anywhere

• Integration of AI tools for code generation and intelligent assistance

• Deployment pipeline for OTA updates to embedded automotive systems

"We can offer a turnkey end-to-end solution running on AWS, catering to the full product lifecycle of automotive systems, as well as similar mission-critical systems across industries we serve," said Avijit Sinha, Wind River president.

Read More Here

🔥 More Notes

• Apple refutes all of the DoJ’s antitrust claims 

• JDK 23: The new features in Java 23

• Lars Bergstrom (Google Director of Engineering): "Rust teams are twice as productive as teams using C++."

Youtube Spotlight

Biggest Scams In Software Engineering

Was this forwarded to you? Sign Up Here