opinion In the Red Hat world, some things remain the same – Fedora is said to remain supported – while others are beginning to surface: AI-driven applications.
An interesting thing happened at this year’s Red Hat Summit. It’s usually a quiet event, but not this time. Coming on the heels of Red Hat’s first layoffs, I expected a low-key show. I was wrong. Rather, at least the energy seemed to build as Red Hat announced the release.
But first, let’s look at classic Red Hat. After the company recently cut its workforce by 4%, many of his Fedora Linux users have found that the popular community Linux distribution has been hit as well. In particular, his Ben Cotton, Fedora program manager, was fired. This has led a Fedora fan to wonder if his favorite Linux distribution will be scaled back.
When I asked Red Hat CEO Matt Hicks that question, he said: “I think there is a great opportunity for Fedora. Fedora remains the distribution base that will determine what Red Hat Enterprise Linux (RHEL) will look like in five years.” and AI will be the default when it comes to driving innovation.That’s why we have a community.Nothing will change for us, but AI is still an important innovation vehicle for us. ‘ added.
When Hicks mentioned AI, he wasn’t just joining a flood of companies “AI-washing” their product lines, much like many “cloud-washed” their products in 2009. . Long before ChatGPT turned AI into the hottest buzzword, Red Hat was working on turning AI into a useful tool.
This started in 2021 with IBM Research’s project CodeNet. Based on this, Red Hat and IBM created Project Wisdom. This allowed users to enter coding commands as simple English sentences. For example, “Deploying a web application stack” or “Installing Nodejs dependencies”.
This grew into Red Hat’s first major AI success, Ansible Lightspeed. It takes the Ansible DevOps program and extends it with IBM Watson Code Assistant. According to Red Hat, this generative AI service enables more consistent, accurate, and faster automation. It uses natural language processing and integrates with Code Assistant to access IBM Foundation Models built on OpenShift, his Kubernetes service at Red Hat.
CTO Chris Wright said: register In an interview, he said that unlike ChatGPT, which built a Large Language Model (LLM) based on essentially any public data it could vacuum, Red Hat’s LLM is curated and domain-specific.
In other words, these LLMs are said to be built on data that Red Hat believes to be correct. Red Hat says that when Lightspeed generates a particular Ansible playbook, a reusable and simple configuration management and multi-machine deployment system, it is based on tested, high-quality data and code. It’s not garbage written by someone in a hurry to meet a deadline.
And it’s not just built on good code. “We embed metrics throughout the end-to-end process, so we make sure the models are accurate,” Wright said. This includes business metrics to ensure that the project is not only technically successful, but also has successful business outcomes.
Another big advantage that IBM and Red Hat bring is that unlike AI projects that grab all the headlines, “we can tell exactly where the data for our domain-specific LLM comes from,” he says. Mr Wright said. This is very different from the response you get when you ask ChatGPT where to get your answers.
This is similar to what the open source community is pushing towards a software bill of materials (SBOM) to ensure that open source code is truly what it says it is. Knowing exactly what is included in an LLM is rapidly becoming a critical issue regarding quality, accuracy and legal issues. For example, if you use code from GitHub CoPilot, do you know if that code is sourced from a copyrighted open source project? Can you be sued if you do? stay tuned. Courts are grappling with just that question.
Businesses will have to address these issues as they emerge from their intoxication with the possibilities of AI.
Red Hat knows that, and says it and IBM are committed to making sure the data LLM is using is correct and legal.
“Businesses need this kind of precision,” says Wright. Furthermore, “When a model is trained on a set of data and the world around us is highly dynamic, the accuracy of the model can fluctuate. To maintain the continuous accuracy of the model, we always You have to put in the effort, and that will form the big picture.” It’s critically important for companies to successfully adopt AI to have a meaningful impact on their business. “
Put it all together and Red Hat, first known as a Linux powerhouse and then a hybrid cloud powerhouse, may well be known as the right AI company as it enters its 30th year. not. . Let’s see. It’s a long way to go. ®
