Will Robots Replace Humans? He Says No!

Original Title: The Human Advantage in the Robotics Revolution
Original Author: Sumir Meghani, Instawork Robotics Labs (IRL)
Translation: Peggy, BlockBeats

Editor's Note: While most people are still debating whether "robots will replace human work," this article argues that humans will not only not be replaced but are instead becoming an indispensable part of the "physical AI system."

The core bottleneck in the current industry lies not in algorithms or hardware but in "data and implementation capabilities." Robots need to be trained by observing skilled humans operating in real environments, but high-quality, diverse physical world data is extremely scarce, leading to the so-called "hundred-thousand-year data gap." This also brings back to the forefront a long-overlooked type of capability—skilled, schedulable, and verifiable human labor.

In this framework, the role of humans is redefined: they are both the "data source" for training machines, providing standardized, annotatable operational processes, and the "on-site nodes" supporting system operations, undertaking maintenance, repair, and remote control. They ultimately enter a "human-robot collaboration market" connected by platforms, becoming a necessary condition for the large-scale implementation of robots.

In fact, technological change will not eliminate labor but will rather restructure labor division. From ATMs to the Internet, every technological leap has been accompanied by job anxiety, but what is often changed is not "whether there are jobs" but "how jobs are redefined." In this technological cycle represented by humanoid robots, the same path is replaying: tasks are broken down, skills are standardized, positions are reorganized, and new occupations emerge.

The real opportunity lies not in "replacing humans" but in who can build the bridge to transform human capabilities into scalable data, operational systems, and collaborative networks.

The following is the original article:

A year ago, I posed a somewhat unusual question for the labor market: What will happen to our platform's "Pros" when robots arrive?

Our vision is to create economic opportunities for Pros and partners worldwide. Today, over ten million Pros rely on us for their livelihood, and many of them have long been pondering the same question. We have a deep responsibility to provide an answer to this.

At the same time, we have also observed an unexpected phenomenon: some robotics companies have started appearing on our application platform, collaborating with our Pros. They need people with professional experience in robot training tasks and access to a variety of business scenarios—which are the environments where robots will be deployed in the future. And what they rely on is the workforce system we have been building.

At that moment, everything suddenly became clear: Instawork can provide human labor for the "Physical AI Economy."

The "Ten Thousand Year Problem"

Ken Goldberg framed this issue as the "ten-thousand-year data gap": on one side, there is a massive amount of data for training language models; on the other side, there is extremely limited and highly specialized data for training robots to perform delicate operations in the physical world.

Note: Ken Goldberg is a highly influential scholar in the field of robotics and artificial intelligence, as well as an artist and interdisciplinary researcher.

It is this gap that has meant that despite billions of dollars flowing into robotics companies, we have yet to see humanoid robots cleaning hotel rooms or unloading trucks in warehouses... at least not yet.

Our estimate is that the entire industry had collected about 100,000 hours of training data by 2024; by 2025, this number grew to 1 million hours; and by 2026, it is projected to reach 20 million hours. This is exponential growth, yet even so, it has only bridged 0.04% of that gap.

More and more companies are joining this race, trying to build humanoid or general-purpose robots: foundational model labs are developing Vision-Language-Action (VLA) models, hardware companies are building physical machines, and players in between are continually emerging. Capital investment has reached hundreds of billions of dollars. And all these participants are facing the same bottleneck: data.

But the key is, we have actually seen this scene before.

When Automated Teller Machines (ATMs) appeared, almost everyone predicted that bank tellers would disappear. But the result was quite the opposite—the number of tellers actually increased. ATMs reduced branch costs, allowing banks to open more branches; and the teller's role shifted from cash handling to customer relationship management.

This pattern has recurred in every major technological transformation: the Industrial Revolution, electrification, the Internet. New technology does not eliminate jobs; it reshapes them and creates more new opportunities.

A new wave is on the horizon, but this time, it looks more like us: with arms, legs, and even eyes.

The Three Acts of Physical AI

Act One: Training Robots

Over the past year, I have proactively reached out to and consulted with some of the best minds in the global field of robotics and machine learning—from researchers and lab directors to entrepreneurs building dexterous robotic hands and full-fledged humanoid robots. They generously shared their time and insights, leaving me impressed. Honestly, we didn't originally belong to this industry; however, the more I listened, the clearer I saw the space where Instawork could make a difference.

One viewpoint was repeatedly mentioned: robots learn by observing skilled humans perform precise physical tasks in a real-world setting. This means that from standardized knife skills for chopping vegetables to navigating crowded warehouses and even tidying hotel beds to brand standards, the challenge lies in the extremely difficult task of collecting high-quality data—you can't just slap a camera on someone and press record. The data must cover diverse environments, tasks, and hand movements; more importantly, the individuals performing these tasks must be truly skilled. Otherwise, a robot trained with "poor knife skills" will only learn "poor knife skills" (which is not good for anyone).

At its core, this is a labor operations problem: how to recruit skilled workers, train them, ensure output quality, and manage a distributed workforce across different regions and scenarios—these are exactly what we have been doing. We have over ten million skill-verified Pros covering hundreds of task types; established deep relationships with partners to access real business settings; and hold data on who can consistently show up, deliver high-quality work. This combination is something no data collection company can replicate from scratch. In fact, many labs have voluntarily approached us, and today we are collaborating with most of the top teams in this field.

Act II: The Rise of Robot "Tamers"

One thing that is often overlooked: robots also need humans.

An executive at a leading robotics company told me that they have a critical component that needs replacement every 4–6 months—a frequency not high enough to justify having full-time technicians but high enough that any downtime results in significant losses. With the proliferation of autonomous driving, delivery robots, and various automation deployments, more and more companies are facing similar challenges: expansion requires on-the-ground support, but having dedicated staff in every market is not economically feasible.

We have already conducted pilot projects with multiple robotics companies, covering services such as battery replacement, part swaps, and robot repairs. At the same time, we have established an hourly worker-focused robot certification system—a first-of-its-kind attempt in the industry. In just the first few weeks, over twenty thousand Pros have been certified.

On the data collection front, certified Pros learn how to operate wearable cameras, capture high-quality videos, annotate sensor data—when a robot lab needs to record hours of bed-making processes in a real hotel suite, they get professionals, not rookies learning on the job. On the technical support side, certified Pros master hardware diagnostics, safety protocols, and maintenance procedures specific to the robot system.

Imagine this scenario: a logistics company deploying an automated robot fleet in over a dozen warehouses. At 2 a.m., a robot in the Memphis warehouse experiences a navigation error, or a sensor module needs to be replaced in a device in Phoenix. Instead of waiting for factory technicians to fly in days later, a certified Instawork Pro can arrive within hours to resolve the issue. Meanwhile, we are also developing VR-based remote control training to support the lab in scaling up data collection beyond the limitations of on-site recording.

If the next decade will see the deployment of billions of AI devices, the opportunity lies not only in maintaining them but in creating entirely new job categories: robot technicians, fleet operators, remote control experts, and even unnamed roles.

Act III: The Market for Human-Robot Collaboration

Last year, I had lunch with the CEO of a global hotel chain. They were seriously considering how to enhance room service consistency through automation. Many robot companies want to deploy products in their hotels, but they struggle to determine what is just "demo-ware" and what is a true "operational outcome." And we are very familiar with these scenarios, processes, and pain points—because we have long been providing services in these venues.

We are building a "robot-as-a-service marketplace"—connecting robot companies with enterprises prepared to automate. We already serve both sides of the supply and demand, which means we are not just "matching," but can truly drive implementation.

The future is not about "robots replacing humans" but about "robots collaborating with humans." This is the goal of the Instawork Robotics Lab: three capabilities, one platform—training robots, supporting their real-world operation, and connecting them to the business scenarios that truly need them.

The Bridge

In every major technological revolution, the question has never been whether new jobs will emerge—the answer is always yes. The real question is: who will build the bridge that connects the present to the future.

We believe that at every stage of this process, skilled humans are needed—from training the first generation of robots to deploying large-scale systems, and designing future human-robot collaboration processes. We hope that Pros on the platform can be involved throughout the entire process.

In the "physical AI revolution," Instawork aims to be that bridge: accumulating deep expertise in the most influential industries; already providing training data for robotics labs; already nurturing certified talent for data collection and on-site operations; and building a marketplace that connects robots with enterprise needs.

We are excited for the next phase.

[Original Article]