A startup will now

@synsation_

296.8K followers · usually 28.2K views a post

Views

87.6K

Likes

4.5K

Comments

30

Shares

182

Selling · SaaS › AI Tool

Scale AIseen in 2 videos →

Why it worked

The video highlights a novel and slightly unsettling way companies are collecting valuable data by incentivizing everyday people to record their tasks, tapping into the curiosity around AI and robotics development.

Summary

Companies are paying people to record themselves doing everyday tasks, like cleaning, to collect data for training AI and robotics. This data is valuable because real-world environments have too many variations to be replicated in a lab, making it difficult for robots to function effectively without extensive real-world training data.

Structure

1Startup offers free cleaning
2Cleaners wear cameras for data collection
3Data sold to robotics companies
4DoorDash pays drivers to film tasks
5MIT researchers explain data needs
6Real-world variations are key

Product placement

DoorDash: Acts - a delivery company that pays drivers to film themselves performing tasks for data collection. Removing it would change the video's topic. Scale AI: Acts - a company that collects and uploads data for AI training. Removing it would change the video's topic. Samsung: Appears - a watermark/logo on a product page. Removing it would not change the video's content.

Original caption

Would you do it?

More from @synsation_

Their other posts in the index, biggest breakout first.

On-screen text

A startup will now completely But when the cleaners shows up, they wear a camera the entire job. And then that footage gets sold Robotics And although this company a lot of publicity, they're not the only ones. now pay their drivers to film themselves washing dishes scale AI has collected thousands of hours of this already. get paid a GoPro on your head and go about your day So why are companies suddenly so interested I actually learned a little bit about why, MIT this last winter. on cutting edge robotics research. And when I asked about all the mainstream that are being marketed right now, They said that we don't have enough data in the physical world, make them useful, like they're being marketed right now. and how your home looks nothing like my home. The dishes are different, There's a fork hidden under the plate. Endless tiny variations that you can't replicate in a lab or a virtual environment. A robot room is going to fall apart in a real environment. And this is why you're to get free things for very, very valuable data. Would you do it though?

Transcript

A startup will now clean your apartment completely for free. But when the cleaners shows up, they wear a camera the entire job. And then that footage gets sold to robotics companies. And although this company is getting a lot of publicity, they're not the only ones. DoorDash will now pay their drivers to film themselves washing dishes and folding laundry. Scale AI has collected thousands of hours of this already. You can now get paid a GoPro on your head and go about your day doing your chores. So why are companies suddenly so interested? I actually learned a little bit about why, when I visited MIT this last winter. They were working on cutting edge robotics research. And when I asked about all the mainstream household robotics that are being marketed right now, they laughed. They said that we don't have enough data in the physical world, to actually make them useful, like they're being marketed right now. Because think of your home. And how your home looks nothing like my home. The dishes are different, the sink is wider. There's a fork hidden under the plate. Endless tiny variations that you can't replicate in a lab or a virtual environment. A robot trained in a perfect room is going to fall apart in a real environment. And this is why you're seeing incentives to get free things for very, very valuable data. Would you do it though?