With all the talk about “artificial intelligence”, it is easy to forget the power of “natural intelligence” and how we can leverage both. In addition, as the old saying goes “two minds think better than one”. If this axiom is true then it follows that “three minds think better than two” and so on. This brings us to another powerful movement that’s powering today’s economy—the power of crowds or “crowdsourcing”.
Crowdsourcing can be defined as “the practice of engaging a ‘crowd’ or group for a common goal”. Crowdsourcing is taking place on with many different activities and across many verticals. Due to our growing connectivity, it is now easier than ever for individuals to collectively collaborate — with their ideas, time, expertise, or funds — to a common goal or cause. This collaborative effort is referred to as crowdsourcing.
One of the earliest and most popular crowdsourcing services is Amazon’s Mechanical Turk, also called MTurk. It utilizes the “human intelligence” of a cadre of individuals who complete small discrete online tasks. These online tasks are things that its requesters or clients, need humans (not computers), to do. MTurk is part of Amazon Web Services (AWS). At first blush, it doesn’t seem like an obvious fit with the rest the AWS Artificial Intelligence and Machine Learning (AI/ML) offerings. But as we’ll see below there is a natural symbiotic relationship between MTurk and other AI/ML services. They can sometimes complement each other as well validate results. Let’s continue by establishing some definitions.
Individuals and businesses that coordinate the use of human intelligence to perform tasks that computers are currently unable to do. Requesters can post jobs known as Human Intelligence Tasks (HIT’s), such as choosing the best among several photographs of a storefront, writing product descriptions, or identifying performers on music CDs.
These are the workers or more colloquially Turkers that can browse among existing jobs and complete them in exchange for a monetary payment set by the requester. Providers at Mechanical Turk accept the HIT’s and are paid small sums for each. Though Amazon is a U.S.-based company, workers (as well as requesters) come from many countries around the globe.
Setting up a HIT
To place jobs, the requesting programs use an open application programming interface (API), or the more limited MTurk Requester site. To submit a request for tasks to be completed through the Amazon Mechanical Turk web site, a requester must provide a billing address in one of around 30 approved countries.
Types of HIT’s
Because of this global diversity, what HITs pay and what they require can vary greatly. Some jobs might pay 1 or 2 cents but take only minutes to complete and require very little expertise of the worker. Other HIT’s require workers to gain qualifications before being allowed to work on them. Qualification may be a test but it could also simply be approval or rejection based on your previous work, location, profile, etc. HIT’s with qualification normally pay more.
- Posting blog comments
- Short editing and writing jobs
- Taking surveys
- Transcription of Documents
- Keyword searches
- Image segmentation and tagging
- The mechanics of Mechanical Turk
MTurk Application Programming Interface
To help developers build and scale their use of Amazon Mechanical Turk (MTurk) AWS offers an Application Programming Interface (API) that can be accessed directly from the Amazon Web Services (AWS) Software Development Kits (SDKs). This means that developers can create and manage HIT’s, create and associate Worker Qualifications, and more in nine programming languages. These include:
- .NET (supporting all .NET compatible languages including C#, F# and many more)
You can also connect to MTurk using the AWS Command Line Interface, allowing you to perform operations from your Windows, Mac or Linux command line and without having to write code. You can learn more about these SDK’s here.