At ScaleHub, we process tens of millions of documents and billions of characters for our customers on a scaled AWS cloud solution annually. Our customers generally face the same challenges: Handling structured and unstructured inbound data always requires some human intervention for data processing and exception handling, including correcting OCR results and fine-tuning AI engines. Using Amazon Mechanical Turk’s platform, we engage thousands of MTurk users around the world daily. Our customers benefit from faster turnaround times, higher quality and an optimized cost structure. We reshape the future of document processing.

As a strategy and innovation group that's constantly inventing new research tools, we love using MTurk to test out prototypes of our new tools. The flexibility of MTurk allows us to quickly try out new things that wouldn't make sense with traditional research panels, such as single question experiments.

The F&B industry has always operated at the mercy of changing tastes and preferences of consumers. Our goal is to surface consumer insights and spot emerging trends, so our clients can effectively respond with effective strategies. Workers on Amazon Mechanical Turk respond to our requests to gather information from menus, websites, and other channels. We are able to leverage these human collective insights to better understand customer needs and uncover important market trends.

At Radiant Solutions, we source trillions of satellite pixels every day, and understanding every object, location, and action on this planet is an enormous challenge. Using Amazon Mechanical Turk's crowdsourcing platform, large communities of users sift through massive volumes of data to tag important objects, features, or locations. These labeled datasets serve as ground truth that helps us train and refine our advanced geospatial algorithms.

Today, brands are capable of producing hundreds of millions of social conversations and stories across the digital media spectrum. For Zignal, natural language processing is critical to rapidly synthesizing this massive amount of media data in real-time. Amazon Mechanical Turk makes it possible to generate human-annotated data for machine learning algorithms quickly and at scale. By harnessing the power of the crowd to obtain high-quality labeled data, we were able to measure and build effective models applicable across the media spectrum.

Snowboy is a highly customizable wake word detection engine that makes it possible for users to pick any wake word they want to call their voice assistant into action. Wake word algorithms are based on neural nets. Usually it takes a well-funded team to recruit the thousands of people needed to provide the voice recordings and additional human training to coach a wake word neural net until it works well. Amazon Mechanical Turk provided us with a low cost, scalable, and global workforce that enables us to generate the diverse training sets required for building such AI models.

At Baidu Research, we aim to revolutionize human-machine interfaces with the latest artificial intelligence techniques. Voice cloning is a highly desired feature for personalized speech interfaces. We introduce a neural voice cloning system that learns to synthesize a person’s voice from only a few audio samples. Besides evaluations by discriminative models, we were able to quickly stress-test the audio samples by crowdsourcing perceptions. Using Amazon Mechanical Turk, we were able to tap on a large number of listeners to rate the quality of the audio and compare it to original human recording.

C-SATS enables surgeons to upload surgical videos for assessment by expert surgeons and reviewers who provide objective and confidential feedback on technical skills. Powered by Amazon Mechanical Turk, this scalable platform will fundamentally change how surgeons learn by giving them the opportunity to anonymously receive input on actual cases to improve their technical skills, which benefits patients, surgeons and health systems.

At AI2, we're pushing the state of the art of Artificial Intelligence, which often requires human-annotated data to train new systems and measure our progress. In particular, we use crowdsourcing platforms such as Amazon Mechanical Turk to build datasets that help our models learn common sense knowledge, which is often necessary to answer basic questions that are easy for humans but still quite hard for machines. Amazon Mechanical Turk provides a flexible platform that enables us to harness human knowledge to advance machine learning research.