In a recent episode of The Wall Street Journal’s “Future of Everything” podcast, Marc Benioff, CEO of Salesforce, shared his outlook on the future of AI, proposing that autonomous agents, rather than large language models (LLMs) like ChatGPT, will be the next big advancement.
“I think we’re reaching the upper limits of LLMs right now,” Benioff stated.
He reflected on the public’s growing fascination with AI tools like ChatGPT, acknowledging that this has led to inflated expectations about the power of AI and an overemphasis on LLMs as the central drivers of progress. Instead, Benioff highlighted the emerging significance of autonomous agents—AI-powered entities capable of performing tasks on their own, such as handling sales communications or managing marketing campaigns. He believes these agents will prove to be more impactful for businesses looking to boost efficiency and revolutionize the way work is done.
Salesforce, already a leader in AI, offers a range of prebuilt and customizable AI agents for clients seeking to automate various tasks like customer service. OpenAI, too, is working toward launching its own autonomous agents, capable of executing tasks like writing code or organizing travel, according to Bloomberg.
Nvidia CEO Jensen Huang has shared similar thoughts, predicting that AI agents and “AI employees” will eventually become integral members of the workforce.
Benioff stated, “We have powerful tools that can drive productivity, improve employee performance, and boost both margins and revenues.” “However, we’re not in the science fiction world just yet—though perhaps one day we will be.”
Benioff pointed out that much of the public’s view of AI agents is shaped by films like Terminator (1984), where a cyborg assassin poses a deadly threat, and Minority Report (2002), in which AI detects crimes before they happen.
While some experts and AI advocates suggest that AI could one day solve complex issues like cancer or climate change, Benioff cautioned that such predictions are premature and misleading. He warned that this overhyping of AI capabilities might distract people from the practical, immediate applications of the technology.
“These AI evangelists who are spreading these exaggerated claims are doing a disservice to customers who could otherwise be using AI to increase their margins, boost revenues, and improve customer relationships today,” Benioff said.
He concluded, “Yes, AI can achieve a lot, but the idea that we’re living in a Minority Report or Terminator world? That’s not where we are now. Maybe one day, but not yet.”
Benioff also referenced WarGames (1983), in which a young hacker triggers a military supercomputer that could lead to a nuclear disaster, humorously noting that he hopes that particular future never comes to pass.
