AI agents are here. How much should we let them do?
Should I create a personal AI agent to help me with my daily tasks?
— Seeking help
As a general rule, I think relying on any kind of automation in your daily life is dangerous when taken to an extreme, and potentially alienating even when used sparingly, especially in personal interactions. Ann I have an agent that organizes my to-do list and collects online links for further reading? Great. An AI agent that automatically texts my parents every week with a quick life update? Terrifying.
However, the strongest argument against incorporating more generative AI tools into your daily life remains impact on the environment these models continue to have during training and generating output. With all that in mind, I dug around The WIRED Archivepublished during the glorious dawn of this mess we call the internet to find more historical context for your question. After doing some searching, I came back convinced that you probably already use AI agents every day.
The idea of AI agents or god forbid “agent AI” is the current buzzword for every tech leader trying to hype their recent investments. But the concept of an automated assistant dedicated to the execution of software tasks is far from a fresh idea. Much of the discourse surrounding “software agents” in the 1990s mirrors the current conversation in Silicon Valley, where tech leaders are now promising a coming stream of AI-powered generative agents trained to do online housework on our behalf .
“One problem I see is that people will question who is responsible for the actions of an agent,” reads a Interview with WIRED with MIT Professor Patty Mays, originally published in 1995. “Especially things like agents taking too much time on a machine or buying something you don’t want on your behalf. Agents will raise many interesting questions, but I am convinced that we will not be able to live without them.
I called Mace in early January to hear how her perspective on AI agents has changed over the years. She’s as optimistic as ever about the potential for personal automation, but she’s convinced that “extremely naïve” engineers aren’t spending enough time dealing with the complexities of human-computer interaction. In fact, she says, their recklessness could spark another AI winter.
“The way these systems are built right now, they’re optimized from a technical standpoint, an engineering standpoint,” she says. “But they’re not at all optimized for human design problems.” She focuses on how AI agents are still easily deceived or resort to biased assumptions despite improvements to underlying models. And misplaced confidence leads users to trust answers generated by AI tools when they shouldn’t.
To better understand other potential pitfalls for personal AI agents, let’s break down the nebulous term into two distinct categories: those that feed you and those that represent you.
Food agents are algorithms with data about your habits and tastes that search through masses of information to find what’s right for you. Sounds familiar, right? Any kind social media a recommendation engine populating a timeline with personalized posts, or a continuous advertising tracking tool showing me those bubblegums for the thousandth time Instagram can be considered a personal AI agent. As another example from the 1990s interview, Mace mentions a newsgathering agent fine-tuned to return the articles he wants. This sounds like my Google News landing page.