The age of Artificial General Intelligence (AGI) – seen as the forthcoming evolutionary advancement in AI – is approaching. AGI denotes a type of AI capable of executing any cognitive task that a human can, representing a considerable progression from “narrow AI,” which is tailored for specific functions. Thus, it encapsulates the anxieties many hold that AI will completely supplant human roles.
As per Jeetu Patel, Cisco’s executive vice president and general manager of security and collaboration, AGI is set to materialize by 2025. During a recent Cisco online media briefing, Patel illustrated a scenario of unmatched technological innovations, societal transformations, and a rethinking of security frameworks to guarantee that the rapid integration of AI does not outstrip its secure application.
“Two varieties of companies will emerge,” he remarked. “Those that are AI-forward and know how to utilize it efficiently, and those that do not, who will fade into obscurity.”
Patel’s insights strike at the essence of the AI-driven metamorphosis: triumph relies on preparedness, not merely adoption. While optimistic about AGI’s potential, he acknowledged the hurdles it presents – notably in terms of safety and security.
The briefing introduced Cisco’s newly launched AI Defense platform, which was unveiled on Wednesday to tackle the unpredictable aspects of AI models. It alleviates risks associated with large AI language models, such as toxicity, “hallucinations,” and susceptibility to cyber threats.
“When a model fails, negative consequences ensue,” Patel noted. He highlighted the disparities in safety procedures enacted by differing model providers, advocating for a standardized method to empower enterprises to innovate without compromising security.
Patel shared with Business Times that the AGI breakthrough anticipated in 2025 will hinge as much on technological acceptance as on possible advancements.
“People swiftly acclimatize to extraordinary innovations,” he stated. “The first time someone rides in a Waymo self-driving vehicle, it seems remarkable. By the third exposure, they grumble about the seating. The same phenomenon will occur with AGI. What is exceptional now will soon be woven into everyday existence.”
Frank Dixon, group vice president of security and trust at the consulting firm International Data Corporation, shared his perspective, equating the swift integration of AGI with the revolutionary effect of the internet.
“The initial interaction with AGI may inspire wonder, but as its assimilation deepens, expectations will escalate,” he remarked. “Progress must align with the diligent effort of practical implementation.”
Patel indicated that security would be critical in unlocking the full potential of AGI. Concerns surrounding safety and security will act as significant impediments to adoption, and AI Defense is focused on addressing these issues through a platform-based methodology, which includes visibility, validation, and runtime enforcement.
This effort is backed by Cisco’s $400 million acquisition of Robust Intelligence, a firm specializing in AI model validation and security. It employs a “Tree of Attacks with Pruning” system, which automates the management of vulnerabilities in AI frameworks.
“Validation processes that previously required seven to ten weeks can now be completed in just 30 seconds,” Patel disclosed.
Dixon commented: “Rather than adding security only after the fact, Cisco is integrating it within the development cycle. This transition transforms security from a barrier into an enabler, allowing developers to innovate without fear.”
Patel contended that every application will ultimately evolve into an AI application, rendering safety and security crucial components of the AI ecosystem. This transition necessitates a new perspective, one that views security as a cornerstone for innovation rather than an encumbrance.
“Just as the software-as-a-service explosion stimulated growth in the security sector, the AI revolution will boost the need for robust safety structures. Our aim is to enable organizations to adopt AI confidently and concentrate on the opportunities it presents.”
At the heart of this initiative lies an intriguing strategy: Cisco is positioning itself as a frontrunner in the journey toward managing AGI while it can still be governed.
As Patel expressed, “We’re advancing quickly, and we must accelerate our pace.”
* Arthur Goldstuck is the CEO of World Wide Worx and the editor-in-chief of Gadget.co.za. Follow him on Bluesky at @art2gee.bsky. social.
- This article was first published in the Sunday Times.