Futurist Amy Webb says large companies are falling behind on artificial intelligence, and much of the money they are spending on the technology is not tied to any plan. Webb, founder and CEO of the consulting firm Future Today Strategy Group, spoke Thursday at the Fortune AIQ Summit in a conversation with Fortune’s AI editor, Jeremy Kahn.

“Let’s be fair, Fortune 500 companies are pretty late to the party on this,” Webb said. “Artificial intelligence didn’t just show up a couple of years ago.”

Webb said she sees “enormous amounts of capital” flowing into pilots “with no strategy ahead of them.” Companies then clash with their own security teams, she said, and end up with orphaned projects that never do anything. In some cases, executives hand the building of what becomes proprietary technology to third parties and then get stuck, while frustrated employees begin building their own tools.

Having AI everywhere means little if workers do not know how to use it, Webb argued. She likened the situation to taking a New York City cab and being handed a phone and asked to type in an address—a pattern she estimated has been constant for about two years. Drivers, she said, have become so reliant on navigation tools that there is “a certain amount of learned helplessness,” much like some leaders in the current AI moment.

Her advice to companies was not to aim to be “AI native,” but to focus on “being flexible.” Even so, she said, firms must build mechanisms to enable that flexibility, and she has not yet seen large companies do so, even as many chief executives work to understand AI. The point carries weight because the alternative, in her telling, is exactly what she described: unplanned pilots, security standoffs, and orphaned projects.

Why gains don’t add up

Part of the reason individual productivity gains fail to compound across a company, Webb said, lies in how firms define return on investment. Most apply AI to the bottom line. The yardstick a company picks helps determine whether gains from individual tools ever compound into company-wide results. She contrasted that approach with Runway, an AI video company whose strategy she described as a clear vision for the future that its tools help advance.

Webb told a story about a friend at a giant company she declined to name. When the firm’s chief technology officer would not approve a secure sandbox, the friend went directly to the CEO, built his own instance, obtained access to a supercomputer, and assembled a team. That group now plans to spend a couple hundred million dollars on AI tokens, Webb said. She blamed the absence of “strong leadership and planning” and said many companies lead with “fear and FOMO.” The story captured the workaround culture she described earlier, in which frustrated employees begin building their own tools when official channels stall.

Runway’s counterpoint

Michelle Kwon, Runway’s chief operating officer, described a company built the opposite way. Runway was founded about nine years ago by three NYU graduates. She said the firm uses AI in “essentially every part of what we do,” and that its entire staff writes code.

Employees have built roughly 215 apps for an internal app store since earlier this year, including people on teams outside engineering, and the company’s newest product emerged from that approach, she said. Runway announced a pilot for an autonomous ad agent the day before the panel. One worker built it in “a handful of weeks,” Kwon said, and it lifted ad output by more than 1,000% from a very small base; the product is now being released publicly—a real-time example of the flexibility Webb said large companies have yet to build mechanisms for.

Runway still buys what is not central to its business, she added, and will not build its own payment system or HR compliance software.

On productivity, Kwon said: “Just because you are using AI, that is not a proxy for your productivity or you doing a good job at work.” She added that AI should be a core part of how people work, as long as it is “responsible” and “doesn’t create work for other people.”

“It’s not helpful if I receive a 100-slide deck,” Kwon said. “What is someone asking me to do with that?” The same goes for chatbot output, which Runway sometimes receives, she noted: people “just copy and paste the output without having thought about, ‘How do I synthesize this?’ [It] isn’t a productive use of anyone’s time.”

Francis Shanahan, chief technology officer of Peloton, offered a comment from the audience about the opposite problem. Employees are “inundated now because they can literally build anything,” he said, and he is finding “burnout to be more and more of an issue.”

Kwon responded that Runway makes a point to “celebrate the wins” and “shout people out.” The company gives its entire staff the same day off, not tied to a holiday, because with summer Fridays, “who’s taking a half day? Who’s taking a full day? What’s hard to track?” When Kahn joked that AI agents could keep working on those days, Kwon replied: “They can work tirelessly.”

Risk and liability

Matt Maher, founder of M7 Innovations and a consortium partner with the MIT Media Lab, asked who is liable when an AI agent breaches a contract or causes other harm.

Webb recounted a coding project she once built that spammed 1,000 people after she deployed it in a live environment with a loop in it. Such failures, she said, are ordinary human errors, not signs of AI “waking up.” Executives, she argued, must “stop anthropomorphizing” AI.

Webb added that security and risk officers should say “Tell me more” before saying no, and that leaders should give them some leeway. She also warned that a wave of devices, such as glasses and charms that listen, will arrive over the next 18 months with little planning. It was the most concrete timeline offered at the panel and turns device governance into a near-term planning question for companies rather than a distant one.

Asked about boards using AI, Webb said her firm has used it for a decade, but the tools are useless if directors do not understand what the data show. She compared the situation to her own cycling sensors. “You can’t disassociate from what’s happening,” she said, “or you are liable.”

For this story, Fortune journalists used generative AI as a research tool, and an editor verified the accuracy of the information before publishing. This story was originally featured on Fortune.com.