Goa, India
(+91) 706 615 3542

Generally Intelligent Secures Cash from OpenAI Vets to Build Capable AI Systems

The private AI market has made strides over the past year, as there has been an influx of investors looking to get involved in the space and startups working hard to build innovative systems that help businesses make smart decisions with their data. One such startup just secured cash from an impressive group of investors to further its efforts. Known as Generally Intelligent (GIVEN), the company aims to build capable AI systems that can be accessed by anyone with an internet connection, which allows businesses and individuals alike to make use of top-tier technology that wasn’t previously available to them. 

Who is Generally Intelligent (GIVEN),? 

Generally Intelligent (GIVEN), is a startup founded by veterans of companies including OpenAI, Tesla and Google Brain. The company uses reinforcement learning and other techniques to train its AIs for certain tasks. Basically has now secured $10 million in Series A funding led by Data Collective, with participation from a16z, Drive Capital, Y Combinator Continuity, Launch Capital and XG Ventures. 

Why Doesn’t Everyone Invest in AI 

There’s a lot of hype around artificial intelligence (AI), but what most people don’t realize is that many companies still don’t understand how it works. Even researchers at Google, who are working on these systems daily, recently admitted that they don’t really understand how it works. This isn’t necessarily a bad thing—to their credit, nobody does—but if you do want to invest in or build an AI-based company then it might not be such a good idea to rely on your own intuition. In order for such projects to succeed, they need engineers who can realistically estimate how far off we are and what needs more research before any investment can be made. 

The Five Pillars of General AI 

In a recent podcast with Y Combinator, Ron Brachman, co-founder of Generally Intelligen t (GI) and former lead on Deep Learning at OpenAI, said that one way in which people think about an artificial general intelligence (AGI) is if you can have an AGI that does everything that humans do. It would be able to read books, understand what it read about them and recall what it has read. It would be able to drive a car and play chess. It would be able to reason its way through things just like we do. If you can imagine it doing all those things then you’ve got an AGI because our brains are certainly capable of all those feats. But there’s another definition of AGI that says: If you can make it better than human level in any one task or narrow set of tasks then it’s also an AGI. For example, AlphaGo Zero was not as good as professional Go players when they started training but after 40 days they had surpassed human level performance. So we could say that AlphaGo Zero is an AGI even though it wasn’t as good as a human being at anything else. This second definition is useful for us because by measuring how well our systems perform against humans we get feedback on how well they’re performing relative to their own previous state and relative to other systems out there. 

The Future of Work and Education 

The ever-present fear of robots and artificial intelligence taking over our jobs has never been more tangible. Recently, DeepMind’s AlphaZero taught itself how to become a world class chess master in just four hours. With these kinds of developments, it’s no wonder why people worry about an uncertain future. But what if we could prepare ourselves now? We’re already on that path with a number of new curricula coming out of Silicon Valley. Generally Intelligent is one such example. Through its curriculum, students learn how to program their own AIs for everyday tasks. They also learn about cutting-edge research being conducted by Stanford and elsewhere in Silicon Valley — all in an effort to set up workers for success in a future where machines will be expected to handle many tasks once performed by humans alone. 

How Do We Get There? 

Creating great artificial intelligence is hard, and it’s only going to get harder. Google, Facebook, and countless startups are making big investments in artificial intelligence, which has sparked a series of high-profile acquisitions. But there’s an obvious problem when it comes to investment in artificial intelligence: The most intelligent people around are working on other problems. The folks behind Generally Intelligent are hoping they can convince some of those top minds in computer science and some members of Alphabet’s open-source research team at DeepMind to leave their cushy jobs and join forces on commercializing A.I. technology. To do that, they need money. So today generally intelligent announced that it has raised $5 million in seed funding led by Founders Fund with participation from Initialized Capital, FF Angel IV, 8VC, and several others. The idea was really just to build a company that could build better machine learning products than what we had access to as consumers or businesses ourselves, says Ryan Orbanek who co-founded generally intelligent with fellow Googler Will Conroy last year. We were building things like recommendation systems or spam classifiers internally but we didn’t have access to state of art machine learning models. 

Share

Leave a Reply

Your email address will not be published. Required fields are marked *