“A lot of companies have plenty of ability to implement AI capabilities in-house if they’re smart about the way they build those capabilities and they’re very careful and conscious about the cost profile of the technologies they put in place,” he says. “There is absolutely a sweet spot of relatively easy-to-access capability at a modest price that many technology organizations are perfectly capable of reaching.”
Liberty Mutual has been experimenting using its own nonpublic, internal version of ChatGPT, called LibertyGPT, for the past two years and has nine use cases in production, including document summarization at scale; 18 use cases in R&D; and a long list of potential uses. Tony Marron, managing director of Liberty IT in Belfast, Northern Ireland, says the nine production use cases save the company about 200,000 hours of human labor and generates about $100 million in savings.
Reaping those benefits required a very high level of change management and integrating business and technology team members, Marron says, including data scientists, engineers, and operational employees. There is no plan for an AGI use case, he maintains.
