AI will not make humans redundant guardrails must anchor its growth says Cisco's Jeetu Patel

New Delhi, Feb 20 (PTI) Artificial intelligence will not render humans obsolete, but it must have strong guardrails with trust embedded at its core, as it can amplify human capability if built responsibly, Jeetu Patel, one of the top AI strategists in the world, says.
    "I reject the notion that the advancement of AI means the extinction of human contribution to society," Patel, the President and Chief Product Officer of Cisco, said on Friday.
    In a wide-ranging conversation with PTI Videos, he said the technology must operate at machine scale while remaining deeply human-centric.
    While consumer-facing tools such as chatbots dominate public imagination, Patel said Cisco's role lies in building the infrastructure that powers the AI ecosystem - high-speed, low-latency networking that connects GPUs, cybersecurity frameworks to prevent misuse, observability tools, and data platforms that enrich AI agents with context.
    "We are the critical infrastructure for the AI era," he said.
    Patel identified three factors that could materially slow AI adoption: insufficient infrastructure, a trust deficit, and a lack of contextual data.
    Globally, he said, there is not enough power, compute capacity, network bandwidth or data centre infrastructure to meet surging AI demand.
    At the same time, trust remains fragile. "If people don't trust these systems, they're not going to use them," he said, stressing the need for security guardrails and model validation mechanisms.
    He also highlighted the growing importance of machine-generated data - such as weather and sensor data - noting that 55 per cent of global data growth now comes from machines, an area supported by Cisco's Splunk platform.
    According to Patel, AI is moving from chatbots that answer questions to autonomous agents capable of taking decisions and actions - a shift that significantly raises the stakes.
    Patel described the industry as entering a second phase of AI evolution, moving beyond chatbots toward autonomous agents capable of taking actions rather than merely answering questions.
    That shift, he said, heightens the importance of guardrails.
    "Imagine me asking the agent to do something, and the agent takes the wrong action. The ripple effects of that on society are far greater," he said.
    Addressing concerns raised by AI leaders, including Dario Amodei of Anthropic, about risks such as deception, misuse and loss of control, Patel said advances in model interpretability and runtime enforcement guardrails are essential.
    Cisco has developed tools to algorithmically test, 'Jailbreak' and apply runtime guardrails to AI models, he said, adding that safety and productivity must advance simultaneously.
    He added that safety and innovation must advance simultaneously. "You cannot treat security and trust as an afterthought. They have to be part of the design construct."
    Patel maintained that AI’s greatest potential lies in combining automation with human judgement.
    "I believe the magic truly happens when you take human instinct and judgement and you combine it with AI scale of automation," he said, predicting that while jobs will be refactored, entirely new industries will emerge.
    He called for sustained public-private collaboration to shape balanced policies that encourage innovation while mitigating risks such as economic disruption, authoritarian misuse and autonomous weapons.
    "We have to build AI to serve humanity," Patel said. "That means that this notion to operate at machine scale with an extreme emphasis on human centricity is something that can't be lost in this whole thing."
    On the India AI Impact Summit, Patel described the event, which drew around 2,50,000 participants, as exceeding expectations and reflecting the scale of the country's AI ambitions.

(This story has not been edited by THE WEEK and is auto-generated from PTI)