Salesforce’s head of AI said that companies are frustrated by generative AI’s unreliability (such as hallucinations) and the AI spitting out inaccurate or biased information, and that these issues are preventing many companies from adopting generative AI in their products.
This leads to broader problems for companies, Salesforce AI CEO Clara Shi said onstage. Fortune “Can you trust AI to drive your business?” was the question posed at the Brainstorm Tech Conference in Park City, Utah, on Monday. For many companies, the answer isn’t that simple. “Companies come to us and want to know how they can deploy these solutions in a way that actually makes a difference,” she explained.
Shi argued that Salesforce is well-positioned to tackle this challenge because customers have entrusted Salesforce with their data and business processes for years. “That provides an ideal foundation for the AI to really inject the context that it needs for the models to actually work,” Shi said.
Shi, who was appointed as the company’s first AI chief just months after OpenAI released ChatGPT in November 2022, said that beyond core trust issues like data security and data privacy, trust is key for all generative AI today. She emphasized that Salesforce implements all of its AI in-house, which helps it understand customer concerns. “We like to drink our own martinis,” Shi joked. “It can be tough sometimes, but I really like the accountability that we have of being customer zero.”
Salesforce has been developing AI to help customers who use the software primarily for customer service, sales, and marketing automation since it released its first Einstein AI product in 2016. Inspired by Salesforce CEO Marc Benioff’s admiration for Albert Einstein, the company focused on predictive AI, which was cutting edge technology at the time.
A few months after the release of OpenAI’s much talked about ChatGPT in November 2022, Salesforce announced EinsteinGPT, one of the first chatbots based on a large-scale language model from a major enterprise. Salesforce has since announced a range of AI products, including Einstein Copilot, Copilot Studio, Prompt Builder, RAG, Hybrid Search, and Einstein Trust Layer.
On Monday, Shi touted a new service initiative called “Einstein Service Agent,” a chatbot designed specifically for customer service reps. The chatbot is trained on a company’s data and can easily hand off tasks to a human customer service agent when necessary. Shi said the chatbot’s focus could expand beyond customer service in the future.
But she added that with these types of products, trust is key, and trust itself depends on the specific context in which the tool is used. For example, she said, “Can I trust an AI to help me figure out which customer to contact for this order?” and “Can I trust an AI to answer customer questions using Salesforce’s new chatbot?” Clearly, at least when it comes to Salesforce, Shih argues, the answer is yes.
CEO Daily provides essential context for the news business leaders need to know. More than 125,000 readers trust CEO Daily every weekday morning for insight on and from the C-level. Subscribe now.
