Examining Moravec’s Paradox with Amazon Go

According to Moravec’s paradox, computers can complete adult-level performance on intelligence tests, but have difficulty performing simple tasks like trying to catch a ball or move around in a space. Their dexterity and ability to perform these tasks is similar to a one-year old child– it’s awkward, wobbly, and usually results in something breaking. The principle states: ,. “High-level reasoning requires very little computation, but low-level sensorimotor skills require enormous computations resources.”1 Moravec made these claims in the 1980s, a few decades after computing was introduced into the workplace and during the introduction of the Internet. Almost forty years later, is Moravec’s paradox still relevant?

First let us look at a simple example found in coding.

In order to make a black square in the center of a screen using JavaScript there needs to be an input of about eleven lines of code that look something like this:

Input:

function setup() {

createCanvas(500,500);
background(“ffffff”);
imageMode (CENTER);
fill(0);
rect(200,200,50,50);

  }

function draw() {

  }

Output:

Screen Shot 2018-02-21 at 3.08.21 PM

To make that square repeat in infinite amount of times takes the same eleven lines of code. Only rearranged in a different order.

Input:

function setup() {

createCanvas(500,500);
background(“ffffff”);

  }

function draw() {

imageMode (CENTER);
fill(0);
rect(random(200),random(200),50,50);

}

Output:

Screen Shot 2018-02-21 at 3.07.30 PM

It would take a significantly shorter amount of time for a human to draw a black square on a piece of paper. However, it would not be possible for a human to just keep drawing squares on that same piece of paper for an infinite amount of time. It only took the computer about a second to almost cover the entire screen with black squares. The computer seems inadequate in the former, but excels on the latter.


A few weeks ago Amazon opened up their first non-cashier grocery store, Amazon Go, in Seattle, Washington. Amazon Go is meant to provide customers with a quick and streamlined shopping experience by removing the check-out component from the grocery shopping process. Upon entering the store, shoppers walk through an electronic check-in terminal to scan their Amazon Go app that is linked to their Amazon Prime account. The shopper does not need their phone out during the shopping process due to multiple sensors and cameras around the store programmed to detect when a shopper has put something in their bag. What is particularly unique about this shopping experience is the lack of check-out lines– both cashier operated and self operating ones. When a shopper leaves the store, they simply walk through a ‘transition area’2 where more sensors add up what is in their bag and charges the user’s Amazon account.

While Amazon Go prides itself on convenience, there are critiques about the lack of human contact and concerns that the store is another threat to low-skilled workers. In reply to these concerns, Amazon Go insists they are not replacing humans with technology, but that human labour is just moved from one area to another. They explain that the employees who would be upfront ringing up customers are now in the back preparing the food you see on the shelves, as well as walking around the floor enhancing the customer experience.3 However, most jobs focused on exchange, like payment for goods and services, or complaints and compensation are left to technology. If there is an issue with your payment or the groceries, customers are pointed towards a kiosk to file and resolve a complaint. This can be extra daunting for Amazon Go customers since the software is still in beta. The software is not programmed to handle random human actions very well. For example, issues arise when a shopper takes something off of a shelf, walks around a bit, decides they do not need it and puts it back on a different shelf.

In the case of Amazon Go, the inconvenience that automation could create while trying to be convenient supports Moravec’s paradox. Technology may outgrow human labour, but can anything live up to a human-to-human exchange? 

 


1. https://en.wikipedia.org/wiki/Moravec%27s_paradox

2. https://www.geekwire.com/2016/amazon-go-works-technology-behind-online-retailers-groundbreaking-new-grocery-store/

3. https://www.geekwire.com/2016/amazon-go-works-technology-behind-online-retailers-groundbreaking-new-grocery-store/

 

Carson Collier

Carson Collier is a second year CCT student interested in User Experience and Project Management. She is the acting Director of Blog at Web services for gnovis and Technical Design Studio Manager. Previously she studied anthropology at Virginia Commonwealth University. During her time at VCU she worked in the Virtual Curation Lab scanning and 3D printing artifacts for digital curation purposes.