Automated Interfaces

Anyone who has ever used a poorly designed interface knows just how frustrating it can be—wasting time, draining energy, and even souring your mood. Whether it’s buying train tickets in a foreign country, searching for museum information, or navigating tasks that require clicking through multiple pages, the process can feel unnecessarily complicated. Now imagine someone could handle that for you. What if that someone wasn’t a person but a computer?

It has been two years since ChatGPT was launched, and millions of people have since used it to ask questions and complete tasks. While the quality of information it provides is remarkable, its greatest value arguably lies in the time it saves. Previously, you’d type your query into Google, hope for relevant results, and then sift through various links, one by one. It was a slow, tedious process that drained cognitive resources. This was simply how things worked: users controlled what to view, website owners monetized their content through ads and promotions, and Google earned revenue from search ads. Everyone, it seemed, benefited from this ecosystem.

But then, OpenAI introduced ChatGPT. By training its model on a vast array of internet content, OpenAI allowed users to bypass traditional search engines and websites entirely. This shift wasn’t without controversy. OpenAI used the robots.txt file—a tool originally designed to guide search engines on indexing content—as a loophole to train its AI. Website owners, who never intended their data to be used in this way, found themselves bypassed. Indeed, had OpenAI openly requested permission to train its model on their content, many likely would have refused. In fact, we rarely stop to ask how content creators currently feel about this arrangement, and it’s hard to imagine they see it as entirely fair.

Now, OpenAI and its competitors are looking ahead for the next chapter in AI history: taking actions on behalf of users. OpenAI’s latest product, Operator, is still in its early stages but promises to browse the web autonomously. It could book concert or plane tickets, order an Uber, or perform other online tasks without requiring your direct input, or at least not all of the time. Using its own internal browser, Operator can function independently of platforms like Google, Apple, or Microsoft. So all you need to do is send a request to OpenAI, and the task is completed for you.

While this sounds promising, it has already lead to resistance. Some companies are blocking Operator’s access to their platforms, fearing it will allow OpenAI yet again to exploit their systems without offering anything in return. Their concerns aren’t unreasonable: automated tools like Operator eliminate user interaction with traditional interfaces, making upselling, ad placement, and brand engagement nearly impossible. This threatens the revenue models of many online services. If enough companies block access, tools like Operator may struggle to deliver on their promise.

In response, websites and web apps may start to demand fees from OpenAI to grant access to their platforms. However, these costs would likely be passed on to users, who would need to pay for products like Operator, among others. This could essentially create a new ecosystem where automation tools rely on licensing agreements, much like the current model for streaming platforms or digital media.

Meanwhile, Apple is pursuing its own vision of automation with Intents, a feature built into iOS. Unlike OpenAI’s open-web approach, Intents focuses on native apps and requires developers to implement specific actions. While this ensures reliability and integration, it may limit how quickly new features become available, as developers need time and resources to adapt. Apple’s collaborative approach could create stronger relationships with creators, potentially giving it a long-term advantage over OpenAI’s broader but more risky strategy.

What does all this mean for the future of websites and user experiences? Well, we may be entering an era where user interfaces take a backseat to APIs—reliable endpoints that allow computers to exchange data. Instead of designing intricate digital experiences, developers might focus on creating systems that can easily be automated by products like Operator. Some companies may even take it a step further and build their own automation models to maintain control and maximize the benefits of this new era of interaction.

For users, the future likely holds a mix of interaction methods. Some tools will still require manual effort, while others will offer streamlined automation. The core value these solutions provide is simple: saving time. For developers and content creators, this shift could bring new opportunities. By offering their services through automation platforms, they may reach broader audiences and negotiate better deals. However, this diversity could also lead to fragmentation, requiring users to juggle multiple automation tools instead of relying on a single ecosystem.

When it comes to generalizing interfaces, I remain skeptical about their current effectiveness. While I can imagine a future where we adapt to this new mode of interaction and developers refine their content to be more accessible, we are still a long way from entrusting an entire process to a computer. The challenge lies in the countless details and nuances that traditional interfaces handle well—elements that abstraction layers, like OpenAI’s Operator, struggle to replicate.

That said, there are undoubtedly low-stakes tasks, such as ordering a pizza, where automation makes perfect sense. For these scenarios, tools like this can truly shine. However, it does make you think: are you really willing to miss out on those extra fries just because you forgot to specify them in your initial request?