AI Talk: Transforming your organisation with GenAI

On 8 October 2024, AI Talk host Kevin Craine was joined by Bryan Phillips, Sr. Manager Pre-Sales Consultant, TELUS; Martn Miller, former Director of AI/ML Production Operations,Levi Strauss & Co.; Richard Rodgers, Founder of  Startup TechLaw;Yemi Olagbaiye, Client Director, Softwire.

Views on news 

A lot has changed since early in 2024, when investors were clamouring to put high valuations on any company with AI technology. Now, as if overnight, Wall Street and others are asking for specifics about ROI, and AI agents will be key for deriving a good return from the investments fuelling next-generation AI solutions and investments. While a common AI chatbot interface is focused on taking a specific user command and getting a response from the LLM, an AI agent will be able to take further actions based upon the results of the LLM response, with or without further human involvement. These composite AI models will combine programmatic or deterministic qualities as well as holistic or unstructured qualities (such as LLM prompts and outputs). 

Only five years ago, a lot of data entry needed to be done in order to anticipate customers’ questions. With the AI agent, it all becomes more predictive. Thanks to the agent collecting enormous amounts of data, errors can be kept at a minimum. However, some people are still holding back as they find that not all the data presented is reliable. The use of AI agents will also require reviews of internal systems and websites to see whether these enable agents to have access to the sources they need. Before jumping into GenAI deployment, a business should consider its strategic direction and how GenAI fits into that. Customer service, sales and marketing are indeed areas where quick wins can be achieved. On the other hand, there is software development and operational efficiencies where GenAI can also move the needle. 

An ethical framework for implementing GenAI 

The technology makes it much easier for marketers to create text and video content. Studios are already investing in capabilities that enable them to multiply a character or animation across different assets. The limitations to this content creation are the potential violations of trademarks and copyrights – which must be looked into before and throughout the whole process. As for the training data of GenAI models, it’s impossible to trace down where it is originating from or who is the owner of the copyright. Another major risk is the lack of transparency. In order to build trust and maintain transparency, it’s still necessary to keep a human in the loop for most of the use cases. In critical sectors such as pharma, it’s key that the GenAI models are trained on proprietary, domain-specific data. Big organisations are increasingly creating AI Data Officer positions, who can look at every touchpoint the organisation has with AI both internally and externally.  Searches with GenAI can be directed into the marketing funnel too the way it was done with browser search, and an AI agent can reach out to the user who made the search. 

Teams that have a knack for experimenting with GenAI to create tools for themselves but don’t know how to manage them can call in professional teams who help them get clear outcomes.  Giving key metrics to the board and explaining them the RoI the new technology can generate can help with winning them over. Risks must always be communicated, though in the context of business impact. Businesses can also talk to software vendors to learn what’s on offer and get more knowledgeable about the technology this way. 

The panel’s advice

• Even with the low-hanging fruits of GenAI deployment, you need to assess what the risks are.
• GenAI should be treated as a junior assistant that knows a lot but isn’t wise yet.
• By introducing accountability mechanisms, you can reduce the risk of AI deployments veering away from company values.
• Mitigate the risks that GenAI presents with the right guardrails across model design, training and testing,
• LinkedIn and Coursera have course offerings that can be leveraged to upskill the workforce.
• Experiment with GenAI to create some pilots and prototypes.
• AI takes the robot out of the human.

For a podcast on AI: open.spotify.com/show/3nE5nRVkQYSvWs8ji7ZKD

For Martin’s e-book, click here.