More than two decades since co-founding what has become one of the world's largest tech companies, Salesforce CEO Marc Benioff says the future is all about the new "AI opportunity."
Benioff founded Salesforce in 1999 with Parker Harris, and they set out to create a cloud-based customer relationship management company, several years before cloud computing would start becoming a more mainstream idea.
The next major wave of innovation at Salesforce will involve artificial intelligence, Benioff said on Monday in his keynote address during the company's annual Dreamforce conference in San Francisco.
"It's squarely about one thing: the AI opportunity," Benioff said, and "it's gonna radically change the landscape... but there's a trust gap."
He continued that a lot of the distrust surrounding AI goes back to how data is collected and used to train foundational AI models, the so-called large language models, that power generative AI tools.
"We're not here to take your data," Benioff said. "Your data is not our product."
Benioff also chatted with OpenAI CEO Sam Altman on Monday about the "future of trusted AI" during a one-on-one panel.
OpenAI's most advanced, publicly available version of its AI model, known as the Generative Pre-trained Transformer 4, or GPT-4, was launched earlier this year. But even its earlier versions were advanced enough to capture the world's imagination and spark a global AI race last year.
The model is far from perfect, though. It's not good at basic math and is even getting worse at performing basic math functions due to a phenomenon known as "drift" when improvements in one area cause other parts to deteriorate.
Another common error observed in generative AI models like OpenAI's GPT-4 is called a hallucination, which happens when the system simply makes things up and spits out incorrect information.
"One of the sort of non-obvious things is that a lot of the value from these systems … is heavily related to the fact they do hallucinate. If you just want to look something up in a database, we already have good stuff for that. But the fact that these AI systems can come up with new ideas, can be creative, that's a lot of the power," Altman said. "You want them to be creative when you want, and factual when you want. And that's what we're working on."
Despite his repeated warnings over the past year or so about the theoretical dangers of an extremely advanced general intelligence system, Altman hasn't seen anything particularly scary in OpenAI's models.
"Honestly, nothing's super scary yet. We know it'll come. We won't be surprised when it does," Altman said, "but at the current model capability levels, nothing's that scary."
In May, Altman addressed Congress during a Senate Judiciary hearing and called for some measure of oversight around the technology.
"If this technology goes wrong, it can go quite wrong. And we want to be vocal about that," Altman told Congress. "We want to work with the government to prevent that from happening."
Altman has also been touring the world to speak with users outside the U.S. and various world leaders about what he considers to be the dangers and opportunities of AI. Over the past several months, he has visited Europe, the Middle East, South Korea, India and Australia. He also visited Indonesia in June, and a few months later, the country issued him a "golden visa."
During a Bloomberg Technology conference held in San Francisco in June, Altman said that "no one person should be trusted" with the development of AI.
In July, the Washington Post reported that the FTC was investigating OpenAI's flagship product, ChatGPT, over potential concerns around consumer harm, data security and privacy.
OpenAI also faces multiple lawsuits regarding alleged copyright infringement. In July, comedian Sarah Silverman joined a pair of class action lawsuits against both OpenAI and Meta. And on Monday, a group of authors sued both companies for allegedly using their work without permission to train their large language models, Reuters reported.
