Operating systems are variously defined in as many shades nuance as there are stars in the sky, because we’re all nerds here — and this is important, stick with me. It can be defined as software that identifies and configures physical and logical devices, or defined as software with assistant, management and monitoring capabilities, or defined as a revolution in the history of computing, abstracting away low-level details to put the power directly in the user’s hands, or any of a dozen other descriptions. Renen Hallak, founder and CEO of VAST Data (and speaker at VB Transform 2024), says AI requires a whole new kind of OS revolution in the enterprise — one that starts with a scalable data platform that can handle vast amounts of structured and unstructured data.
It’s also the foundation of every type of computing, past present and future, however nitty gritty you get with your definitions.
Back in the great wild west of computing history where punch cards roamed free, every vendor was pretty much making their very own bespoke handcrafted artisan operating systems for their own private mainframe computers, and every one of them was like a beautiful snowflake, no two models of commands, operating procedures and functions alike, and things quickly got complicated in the OS world for awhile.
Today, however, it’s mostly just a handful of dominant players. The new question, as generative AI takes over all of our waking thoughts, hopes and dreams as well as our enterprise infrastructure, is what does the operating system that can handle generative AI look like, and will there ever be just one?
Something extra special, for sure. The demands of an operating system that enables gen AI to be its best self are many and specific. It includes some very advanced dynamic resource management, real-time processing, new kinds of security and a sharp eye on ethical use, support for edge computing and a layer of abstraction to allow algorithms to run seamlessly across a variety of hardware architectures. It’s got to have middleware and framework support, all-important scalability and distributed computing, and it probably shouldn’t be a climate killer, which means some major power management in there too.
But it’s also got to be infrastructure that can handle the multifaceted maze of interactions between generative AI applications and the varied other enterprise assets that make up an organization’s technology stack. LLMs are out here generating content and making autonomous decisions that ripple through the entire organization, and they need support to coordinate their functions, deliver resources where they’re needed. It’s not about adding another layer of frosting on a day-old cake, it’s about AI managing AI, and giving humans the power they need to use that AI.
In other words, something that will be a revolution in the history of generative AI by abstracting away the low-level details and putting its power in the user’s hands. But meeting the needs of AI while not kicking less cool applications out of bed can be a major challenge — the battle between specialization and flexibility is eternal and ongoing.
Intuit recently took a page out of history and created its own proprietary OS for AI, called GenOS. But that’s internal software. An AIOS that hits the kind of ubiquity of a Windows or a Linux, standardized and broadly compatible will be hard to build, considering the diverse and densely populated AI hardware and software ecosystems — or will it be?
Hallak has some more thoughts about that. At VB Transform 2024 he’ll be exploring the idea of an AI OS that will redefine how generative AI is used, and dive into what it will look like, and why its got to start with a scalable data platform that can handle the wild amount of structured and unstructured data that modern AI applications need. He’ll also be talking about the strategic considerations for building scalable, cost-effective and future-proof AI infrastructure for cutting-edge innovation, and sending attendees off with new ideas about architecting AI frameworks, data management and processing, and more.
Don’t miss this session and more, plus plenty of gossip, networking and industry-changing happening live in San Francisco at VB Transform 2024, July 9, 10 and 11. This year the theme is putting AI to work at scale, with a focus on the practical gen AI case studies and application stories that matter most, directly from the industry leaders with the biggest brains. Register now!
Source link
lol
