Introduction: A Three-Course Meal for the Right to Film Your Kitchen
According to an article by Reece Rogers in WIRED, when the private chef sent by the startup Shift arrived at his apartment to cook a three-course meal, he was already fully prepared to record. The chef walked through the front door wearing a baseball cap with a blocky, white camera mounted on the brim. Rogers had signed up for a free lunch in exchange for agreeing that the entire experience would be recorded and uploaded as training data. As the chef, Ollie, unpacked his equipment and began preparing a gazpacho soup, the hat-mounted camera recorded his hand movements with high precision. The goal of this recording is to help build a better robot in the future—one with high culinary expertise.
This is part of a growing trend of egocentric data collection—recording human hand movements from a first-person point of view. Shift made waves earlier this year by offering free cleaning services in New York City under a similar arrangement: allow Shift to record the execution of various chores in your home, and in return, you receive the service free of charge. Now, the company has expanded its operations to San Francisco by offering free home-cooked meals. If in the not-so-distant future we are able to purchase a skilled humanoid robot that can live alongside us at home, we will need to ensure that its knife-wielding skills meet the required standards.
The Vision Behind the Venture: Shift and its German Parent Company Microagi
Shift is a division within the larger German startup Microagi, which focuses on deploying robots in the real world and human living environments. Microagi's CEO, Bercan Kilic, explains that Shift's goal is twofold. First, to allow ordinary people to join the AI economy and sell their data. The second part, he says, is to handhold the entire society and governments together, from today until the era of economic and technological abundance.
In the short term, Kilic sees this initiative as a way to build a marketplace where independent contractors can earn money by recording videos and contributing to the improvement of AI models. Looking at the further horizon, he believes that the impending technological innovation could usher in a new, utopian-esque era of widespread, shared wealth for everyone. However, the road there involves collecting a massive amount of valuable physical data.
The Gig Economy of Body Movement Recording
Rogers describes how he spent dozens of hours over the past year with an iPhone strapped to his head, trying to understand this emerging sector of the gig economy. He recorded his hand movements for DoorDash while scrambling eggs for breakfast, and repeatedly recorded his fingers tying shoelaces for another data collection startup. Shift offers a similar contract-work setup for people who want to record and sell their first-person vision data.
Although the payouts Rogers received for these tasks were quite meager, he had previously managed to generate some income from his physical movements on other platforms. This experience sparked his curiosity to discover what it feels like on the other side of the camera—to be the one being recorded in his own home for the purpose of collecting data intended to train the robots of tomorrow.
What Happened in the Kitchen: Precise Cooking and Meticulous Cleaning
The uniforms of the Shift chefs combine elements of restaurant workwear with features that look almost cyborg-like. Chef Ollie wore a white shirt and pants with a brown apron wrapped around his body. The camera attached to his hat was roughly the size of a GoPro camera, with a white wire running from its top and down his back. The wire was connected to a smartphone placed in his pocket or on his body, where the recordings from that free lunch were stored. The camera jiggled slightly while he cooked and prepared the dishes.
Rogers, who had never hosted a private chef in his cramped apartment, describes a slight sense of strangeness standing next to the chef in the kitchen while he prepared the gazpacho appetizer and began chopping trumpet mushrooms. The finesse and dexterity shown by Chef Ollie in his knife work—which was recorded on camera—far exceeded Rogers' own rudimentary skills, which could not provide high-quality training data for a future robot. The journalist's sense of awkwardness grew when he remembered everything was being recorded, and he thought he should have scrubbed the oven better before it was immortalized as robot training data. The kitchen filled with wonderful aromas as Ollie began preparing the main course—baked salmon in a creamy zucchini sauce.
After finishing the main course, the chef completed the meal with a beautifully decorated tiramisu dessert and immediately began cleaning the kitchen. Rogers notes that collecting cleaning data is no less critical than collecting cooking data, as a humanoid robot would not be very useful if it cooked a Michelin-star-level meal but left a pile of dirty pans in the sink.
The Need for Unique Data and Comparison to the AI Revolution
According to Microagi CEO Bercan Kilic, Shift plans to expand in the future and add more free services, or at least offer them at discounted rates, in exchange for recording the experience. While the breakthrough of large language models (LLMs) was driven by scraping the entire internet to extract text data, there is no comparable massive repository of egocentric video data for training physical robots. This is why the inside of your home, along with the way you navigate it and complete tasks, is so valuable that a company is willing to fund a private chef for you for an entire day.
Kilic expresses great optimism about the near future, claiming we will soon live alongside humanoid robots. "By this time next year we will have quite good home robots that are at reasonable prices," he says. "They won't do all the tasks, but they will do most tasks quite well. It will be like vacuum robots in 2020."
Safety Concerns and Socioeconomic Skepticism
Alongside the technological enthusiasm, Rogers raises concerns about the physical implications of giving tools and physical control systems to AI models. While a vacuum robot (like Roomba) might at worst smear dog poop on the carpet when it malfunctions while the owner is at work, a novice chef robot holding a sharp knife could endanger pets or humans themselves if a glitch occurs in its system.
Furthermore, the journalist expresses skepticism about the promise that these robots will lead to an era of economic abundance for the general population. He mentions that while the Industrial Revolution did change many aspects of the world for the better, wealth inequality is worse today than ever, and homelessness in the United States remains close to all-time highs. In his opinion, the robotics revolution has the potential to improve basic living conditions for some people, but it will not necessarily eliminate poverty completely. The early, affluent adopters who are willing to spend thousands of dollars next year on home robots will have to partially thank Rogers' apartment kitchen and Chef Ollie for the meals those robots will prepare for them in the future.