Artificial intelligence has deep historical roots dating back several decades, and in the early days, the focus was on improving accessibility, enhancing safety and bridging cultural divides.

The breakthroughs we’re seeing in the field today owe their existence to the many individuals who have worked to improve it over time, according to Igor Jablokov (pictured), chief executive officer of Pryon Inc.

“If you think about your texting while driving scenarios … with early machine translation and the like, many of us have been working on the soup for a long while,” he noted.

Jablokov spoke with theCUBE industry analyst Rob Strechay at Supercloud 4, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the journey of AI from its inception to its current application in critical infrastructure and knowledge management.

Knowledge management and accessibility

Pryon brands itself as a secure, responsive and trusted AI knowledge fabric weaved from an organization’s unstructured content, empowering users to unlock any answers they need. The company, in a nutshell, is dedicated to making knowledge management more accessible.

Its mission is to connect and organize disparate sources of information, including documents stored in various locations such as cloud drives and drop boxes.

“The first step, when we got started many years ago, was there was just a dearth of enterprise AI,” Jablokov explained. “There was just nothing in our respective workplaces. It didn’t matter if you were attending a manufacturing plant or a hospital, or you were in a government agency. There was just no enterprise AI to speak of over half a decade ago, so we decided to catch our own football.”

The idea behind Pryon didn’t spontaneously manifest either. It’s a result of Jablokov’s years of involvement in the early iterations of products such as Siri and Alexa.

Challenges for new entrants

With any new technology, organizations are bound to face certain operational hurdles — especially as they navigate the nuances of directing the tech toward delivering value. Many organizations are now exploring AI, trying to figure out where to begin and how to approach it.

The paradox, however, is that pre-existing consumer AI models — such as in streaming boxes and smart speakers — come with pre-loaded models ready to use, while enterprise AI often requires starting from scratch, according to Jablokov.

“The paradox to bringing this into your workplaces is that I would have to show up with a blind trust, essentially an empty vessel that has no idea what content you’re going to put into it,” he said. “And yet you’re going to be expecting top marks on accuracy, scale, security and speed — especially if you’re going to be introducing it into critical infrastructure.”

Customization and segmentation

Organizations define efficiency and value in vastly different ways. Thus, reaping the benefits of AI largely depends on customizing the technology to suit particular use cases. But there are risks to large-scale large language model adoption, according to Jablocov.

“Unfortunately there are about 17 different risks associated with the adoption of LLM-style technologies in these enterprises — Pryon was actually designed differently,” he said. We took a third approach, which is to only leverage these styles of technologies to model language but keep the enterprise’s content from not mingling with that core itself. And that way, you don’t have the issues, from model drifting to hallucinations, that you’ve seen discussed all over the press.”

This approach allows organizations to maintain control over their data and content, ensuring that AI-generated responses align with their requirements and business objectives. It’s a method that strikes a balance between rapid deployment and customization.

Privacy and security

Security is a top priority for organizations, especially when dealing with sensitive and critical information. Pryon’s approach to security involves running the platform on resources that organizations can control, integrating with existing security constructs such as single sign-on and providing fine-grained user access to knowledge collections.

“It means allocating the right users to the right knowledge collections,” Jablokov said. “It means presenting the answer only with certain contexts. You’re in a certain location, and you’re allowed to see certain things. It means whatever document-level controls that you have are preserved by the system. When it gets ingested into the system, you’re also preventing spills.”

Every answer generated by the Pryon platform is anchored to an organization’s fundamental assets, making it transparent and traceable, thus organizations can be confident that the information they rely on is trustworthy, Jablokov added.

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Supercloud 4:

Photo: SiliconANGLE