Since content moderation continues to be a critical aspect of how social media platforms work – one that may put pressure on them to get the right results, or at least get better at tackling them – a startup has a number of data and image models created to support this Along with any other task that requires automatic recognition of objects or text, a large round of funding is announced.
Hive has built a training database based on crowdsourced contributions from around 2 million people worldwide and then supports a number of APIs that can automatically identify images of objects, words and phrases – a process that is not just used to moderate content Platforms, but also in creating algorithms for autonomous systems, back office computing, and more – raised $ 85 million in a Series D funding round, which the startup has confirmed is worth $ 2 billion Have US dollars.
“Our work is focused on developing AI models that can help automate manual work in the past,” said Kevin Guo, co-founder and CEO of Hive. “We’ve heard of RPA and other workflow automation, and that’s important too, but it also showed that there are certain things that people shouldn’t do that are very structural, but these systems actually can’t address a lot other work that is unstructured. “Hive’s models help improve the structure of this other work, and Guo claims they offer” near-human accuracy. “
The funding is being led by Glynn Capital, which includes General Catalyst, Tomales Bay Capital, Jericho Capital, and Bain & Company, as well as other unnamed investors. The company has now raised $ 121 million, making this final round an especially big leap.
The company has been a little under the radar since its inception in 2017, which apparently was a linchpin of founder Kevin Guo’s previous startup, a Q&A platform called Kiwi, which was itself a product of a project from its time in Stanford. But since then it has quietly taken on some interesting clients including Reddit, Yubo, Chatroulette, Omegle and Tango, along with NBCUniversal, Interpublic Group, Walmart, Visa, Anheuser-Busch InBev and others. In total, it has around 100 customers and has grown by more than 300% in the last year.
Hive began by identifying images and working with companies building autonomous systems. When you talk to Guo about Zoom, you will likely get a screenshot of some of this work as a backdrop as cars race across the Golden Gate Bridge.
Today, most of Hive’s activity (excuse the pun) revolves around moderation, some of which contain images, others text and streamed audio – which is converted to text and then moderated as it would be. (I believe autonomous car modeling is still used as a background as it is a little less disruptive than a content moderation image as you can see below.)
Firstly because it’s a very classic problem that you can imagine being solved or aided by the use of AI, and secondly because it’s such a big problem on the internet today, there is one Range of other startups creating platforms to manage online abuse, including harassment, and help with content moderation.
These include Sentropy, Block Party, L1ght, and Spectrum Labs, not to mention lots of tools built by big tech companies themselves. (Instagram, for example, just launched its latest tools today that users can use to combat abuse in DMs: it created the whole thing in-house, the company told me.)
But as Kevin Guo describes it, what made Hive stand out from the crowd was the crowd, so to speak. Over the past few years, the company has slowly built up a database of “abuse” or “abuse” through crowdsourcing feedback from around 2 million users who are paid with either “normal” money or Bitcoin to go through a series of images and text elements in sequence identify other things. (Bitcoin started as a side offer and now makes up the bulk of contributor pay, Guo said.)
This database, in turn, supports a number of APIs used by Hive customers to aid them in running their own moderation tools, or for any workflow that requires frequent and quick identification.
Most of the language learning in the system is currently based on English and several other popular global languages such as Spanish and French. Part of the funds will be used to expand reach and global coverage, also in a broader usage. It also leads to a wider range of use cases for the data and technology created by Hive.
One of them, Guo said, involves a new approach to advertising based on showing ads related to something you may have just read or seen on screen. Very GDPR friendly as it involves absolutely no involvement of any data or your online browsing activities (anonymized or not). This is gaining traction with brands that originally came to Hive to help protect their IP or reputation management and these are now how they can use the tool to get the word out about themselves more effectively.
The ways Hive’s AI can be used in the future is part of what attracted the investment today. The focus on how it was created in the cloud underscores this extensibility.
“Cloud computing has seen tremendous adoption in recent years, but only a small fraction of companies are currently using cloud-based machine learning solutions,” said Charlie Friedland, principal at Glynn Capital, in a statement. “We believe that cloud-hosted machine learning models will be a key component of cloud growth in the years to come, and Hive is well positioned as an early leader in this space.”