AI start-up tackles information overload with automated video summaries
11 April 2022
A start-up based at CeADAR, Ireland’s Centre for Applied AI, is developing technology that can tackle information overload and shortening attention spans by reducing the length of recorded videos.
Giyst uses Artificial Intelligence and machine learning to create summarised videos for business, education and other markets.
The summary generated by giyst is a fully-edited automated video that can transform a 50-minute lecture, speech or training module into a five-minute video of the essential information and provide a transcript.
Video summarisation as a service is considered to have huge potential across a number of sectors as companies increasingly use videos for their sales, marketing, events, communication and training programmes.
Giyst is aiming to transform the global enterprise video platform (EVP) market by enhancing user experience through faster content discovery and allowing people to get to the point as quickly as possible.
The EVP market is estimated to be worth $25.60 billion globally by 2025. Video is also experiencing huge growth as a tool to enhance the learning process in the education sector. But consumers of video content in business, education and other sectors are struggling to keep up with the number of videos they are expected to watch for work or education purposes.
Giyst identified information overload as a critical daily issue for businesses, agencies and consumers. Currently, 82% of all created content is in video form but only 24% of people will get to the end of a video if it is over 20 minutes long.
In addition, workers are wasting time by spending 2.5 hours a day trying, and often failing, to find the right content, and shortening attention spans are affecting our ability to consume information. (*source of stats in notes below)
The team in CeADAR comprises co-founders Colm Murphy and Avril Power, who are in charge of business and commercial development, Sean Mahon, data scientist with expertise in natural language processing, Hanene Jemoui, who is the software architect of the project, and Dr. Ricardo Simon Carbajo, who is the lead of the project and co-founder of giyst.
The giyst team is now focused on further enhancing the summarisation engine and converting current trial partners to customers, with a view to going for seed funding by Q4 this year.
Dr. Carbajo, head of innovation & development at CeADAR and giyst project lead (pictured), said: “The giyst start-up project addresses one of the main challenges when consuming digital content, i.e. how to get the gist of a video in such a short time that suits our daily digital content consumption requirements and constraints.
“Nowadays, the latest advances in deep learning, specially the so-called transformers, are enabling us to train and fine-tune customised models for natural language understanding and summarisation; this technology is at the core of giyst.”
Avril Power, Co-Founder of giyst, said: “We can’t keep up with the amount of video we are required to watch, and we can’t keep tuned in, because we haven’t got the attention span. On top of that, we also can’t find the right content because of information overload. You might start watching a video that you think is valuable and then you realise it isn’t the topic that you wanted.
“By providing instant, automated video summaries of the essence of content, giyst drives four times more content discovery and maximises audience retention by at least 40% thereby creating efficiencies for consumers and the businesses that provide them with video content.”
This content was originally published here.