Artificial intelligence (AI) is transforming the technological landscape of the future. US-based technology company, is developing a consented and privacy preserving public dataset of human interactions for AI research. This video content will help researchers to advance the possibilities of machine learning within their applications. The contributors to the project are remote workers in the gig economy, also known as the ‘human cloud.’ They are being crowd-sourced by freelance hiring platform, Twine, through their extensive community of 350,000 freelancers.
Twine’s freelancers will create videos of everyday activities to provide detailed, high-quality examples of human interactions, involving everything from getting into cars to exiting rooms, which will assist AI machines in learning to replicate these actions more effectively. Many AI research projects are forced to rely on content from YouTube, which has held back the development of artificial intelligence as it does not have the quality of material required. Twine’s freelancers are helping to build a dedicated dataset of human interactions which will help to advance the development of learning algorithms.
The video library will also be provided to the international research community as an open source dataset, where contributors have given explicit permission, so that researchers can benefit further from the materials to advance machine learning. They are focused on producing this dataset in the most ethical and effective way possible by paying freelancers for each video they create. This is helping to plug the gap due to the lack of demand for freelancers during the Coronavirus outbreak.
Twine is a marketplace that connects companies to a network of over 350,000 creative and digital freelancers. Twine is therefore perfectly set up to provide the subjects and creators for large scale video AI research projects. As a Twine Enterprise customer, the US research company is receiving a fully managed service from Twine, meaning that all the contributors are sourced from the Twine network and managed by members of the Twine team so that they can focus on the outcomes of the research.
Stuart Logan (Twine’s CEO) says:
“74% of Twine’s freelancers are more likely to remote work in future as a result of the COVID-19 pandemic. This was already a growing trend before due to the improvement of communication tech. The pandemic is only going to accelerate the trend. If I was an office-space landlord, I’d be worried.”
“I’m proud we’re able to use our global network of 350,000 freelancers to deliver a unique project that is at the forefront of AI by fully remote freelancers.
“This partnership gives Twine a huge opportunity to grow in the world of AI and demonstrate how mass scale remote working can be a success. One of the main challenges of machine learning projects is they need vast amounts of training data’ and with Twine’s network, we can help companies get that. Remotely.”
With over 300 freelancers currently participating in the project, they have already got off to a great start and this will continue to grow. The project is running until October and is expected to have had over 3000 creatives involved within this time period.
The Twine survey had 256 responses. “Are you more likely or less likely to work remotely in future?” 74% answered “More likely”
View www.twine.fm to find out more.