The topic of generative AI was top of mind for attendees and executives at Cisco Live in Las Vegas last week.

Cisco executives were quick to point out it has worked with so-called hyperscale companies on the technology and implanted it across its portfolio, but generative AI poses new challenges, especially for a company with large amounts of customer data.

Here’s a look at six key topics for the use of generative AI.

Threats
During an executive panel, Cisco EVP and GM for security and collaboration Jeetu Patel summed up the risks and rewards of using generative AI.

“I’ve never been more pumped about a transition and I’ve never been more paranoid about what could go wrong.”

“We have to make sure that we are very careful and not just go up haphazardly.”

He stated part of Cisco’s approach to generative AI includes working with the public sector to ensure a long-term regulatory framework is put in place, because “it’s going to be a fundamentally game-changing era”.

Ethics
Dev Stahlkopf, EVP and chief legal officer, has been tasked with creating a responsible AI framework for Cisco, which she stated includes slowly unlocking tools and putting protection around data sets.

She noted the vendor’s customers have asked how much they should trust generative AI and how Cisco is integrating it with its tools. She stated it all comes down to transparency and accountability during conversations with customers regarding where and how the technology is being used, and if training is taking place on synthetic or purchased data.

“We’re not using customer data.”

“I think we’re also very careful that the use case fits the purpose because it can overstep quickly.”

IP protection
Speaking on the sidelines of Cisco Live, David Burns, AVP of network automation and shared services at Cox Communications, said one of his issues over the use of generative AI chatbots is a lack of clarity about which source code or open language is being used as information is gathered.

“Where we have concerns is IP”.

“When do the lawyers get involved? We’ve got to be very careful and be very cautious of that.”

Burns expressed optimism AI will boost the work of software developers by providing the tools to help them “annotate and document their code”, an area he noted is often a weakness in the sector.

Cost
Cisco CEO Chuck Robbins said it is too early to talk about capex for the company’s plans for generative AI across its portfolio.

“I would say that we’ve run the whole gamut of what possibilities are out there and we haven’t come to any final solutions yet.”

Modelling
While the vendor is gauging how to deeply ingrain generative AI across its portfolio, Cisco unveiled several products based on the technology.

Patel stated current opportunities include data, foundation models and better employee or user experiences.

“We will create specialised models and we will make sure that we stay on top of the foundation models,” he stated, adding Cisco plans to “make sure that we’ve actually worked with use cases where the data is unique to us”.

Framework
Chief strategy officer and GM for applications Liz Centoni said the telecoms industry and countries around the world need to understand the complications of using generative AI, which includes establishing a consistent framework.

Stahlkopf argued regulations should include two elements: a risk-based approach for the use of AI to establish the biggest threats; and engaging in talks across jurisdictions to prevent patchwork regulations which are ineffective and limit innovation.