Twitter has acquired a London-based startup, Magic Pony, in a bid to ramp up its artificial intelligence (AI) and machine learning capabilities, in a deal reportedly worth $150m (£102m).
"It’s powering much of the work we’re doing to make it easier to create, share, and discover the very best content so that every time you open Twitter you’re immersed in the most relevant news, stories, and events for you."
In a note on Magic Pony's website, the company said:
"Together with Twitter, we're looking forward to vastly accelerating our rate of research; growing our team, which will serve as the European homebase for Twitter's machine learning efforts; and continuing to publish. We've barely scratched the surface of what we think is possible in this area and are excited to see what the future holds."
It's not the first time a US tech firm has come calling on the UK's expertise in artificial intelligence. Google snapped up Deep Mind in 2014 for £400m.
The race towards artificial intelligence is largely occurring through machine learning, where computers are able to teach themselves how to perform functions without input from a human programmer. Deep Mind's AlphaGo recently beat the world champion player of the chess-like game Go with machine learning.
Dorsey said the acquisition builds upon previous AI and machine learning companies it has bought, including Madbits in July 2014 and Whetlab last year.
Magic Pony was founded in 2014 by entrepreneurs Rob Bishop and Zehan Wang, both engineers who attended Imperial College in London, and is developing machine learning software for visual processing. It landed investment from Octopus Ventures, Balderton Capital, and Entrepreneur First at seed stage.
It's 14 strong team, which are largely PHDs from some of the world's top research institutions, will become part of Twitter's Cortex engineering team.
Balderton partner Suranga Chandratillake said of the invesment in the company:
"Looking back, deciding to invest was easy. The founders had three key ingredients as a team: a consistent, strong vision of how they could build a very large company, the real-world experience to build an actual business; and significant technological depth that came from being experts in their fields. I also felt confident that we could help the company.
Most likely large customers were based in the Bay Area (where I spent over a decade) and Rob and Zehan’s primary focus was on developing technologies for video processing and, as the founder and chief executive of a video company for eight years, I immediately could see the impact that their technology could have."
In an interview with MIT Tech Review earlier this year, the company spoke about how the technology works and its implications.
Via machine learning, a computer is able to fill in the detail of low resolution images and video using minimal processing power. Embedded in everyday computers and smartphones, the technology could improve the photos we take or the video we're streaming.
Bishop said: “Online video-streaming businesses rely heavily on video compression. Our first product demonstrates that image quality can be greatly enhanced using deep learning, and fast mobile GPUs now allow us to deploy it anywhere.”
With video parts of the Twitter business such as Periscope and Vine, the integration of Gifs earlier this year and the visual nature of the platform in general, the technology could greatly increase the quality of images and video without being slowed down or costing more of users data.