What You Need to Know About 'AI Agents' and Why We Are One Step Closer to The Jetsons This generative new technology will be taking over a lot of busy work, imminently.

People have wanted robot assistants ever since...The Jetsons? Now generative AI seems ready to deliver. And these assistants have a name: They're called "AI agents."

Unlike a chatbot that can only compile a list of curated information — like naming a bunch of hotels within your budget in Paris — an AI agent would be able to act on your behalf. It would book a room in one of those hotels, plus a plane ticket to get there. It would negotiate a refund if you needed to cancel, allowing you to bypass the hassle of looking up cancellation policies and waiting on hold. Based on your travel schedule, it would also know how crowded or popular a certain area is, and could make recommendations using that knowledge.

"Software is going to change from something you use to get a job done to something that can get the job done for you," says Clay Bavor, cofounder of AI startup Sierra, which provides companies with the tools to create their own AI agents.

Related: AI Is My Meal Planner, Editor, Assistant and More as a Working Mom — Here's How You Can Use AI to Conquer Multitasking, Too.

Now, don't cut "human assistant" from your budget just yet — we've got a few more years of development. But Bill Gates expects this technology to become widespread within five years. "In the near future, anyone who's online will be able to have a personal assistant powered by artificial intelligence that's far beyond today's technology," he wrote in a November blog post. He also thinks AI agents can streamline and democratize access to certain services, such as education or healthcare, by providing simple, personalized access in one place.

"It's inevitable that this technology — or some evolution of it — is going to be deployed at a massive scale," says Zach Koch, cofounder and CEO of Fixie, a startup focused on teaching AI-powered agents to communicate verbally in real time rather than through text.

Of course, AI agents, like other forms of AI, come with ethical considerations. To provide customized, nuanced service, an AI agent must know a lot about its user. It needs to plug into many data streams, which could range from Slack channels and email to medical files and one-on-one conversation transcripts, depending on the service. This has people asking: Who owns the data a user shares with the agent? Can law enforcement use that data? Can the agent refuse to do something harmful?

Indeed, the increasing autonomy of AI caused the Center for AI Safety, a nonprofit research group, to list AI agents as an example of a "catastrophic AI risk," writing that "malicious actors could intentionally create rogue AIs with dangerous goals."

Still, many in the space argue that the tech cannot do its job without access to sensitive data. They also say they're already accounting for such potential risks.

"On data, our view is simple," says Bavor. "Our customers own their data, and we've built strict data governance into the foundation Sierra platform." He explains that an AI agent should be taught to transfer risky conversations — for example, someone expressing intent to harm themselves — to a human for help.

Want to experience the future now? You can, sort of: There are AI agents out there, but they're expensive and relatively unreliable. "I think we're a few major research innovations away from seeing real progress here," Koch says, "We've gotten better at understanding what it takes, but it's far from a solved problem, or even fully understood." Which means that, for now at least, your best assistant is still a human.

Related: Chatbots vs. Virtual Assistants — Which is Better? A Complete Guide For Your Business Needs