Elon Musk's non-profit OpenAI wants to build a household robot and AI agents

Tesla CEO Elon Musk's $1bn non-profit artificial intelligence (AI) research firm OpenAI has announced its major technological goals which include creating a household robot, a natural language processing chatbot and an intelligent agent capable of winning any game. Announced in a blog post on 20 June, the firm says "robotics is a good testbed for many challenges in AI".

"OpenAI's mission is to build safe AI, and ensure AI's benefits are as widely and evenly distributed as possible," the company says in the blog post authored by Ilya Sutskever, Greg Brockman, Sam Altman, and Elon Musk. "We're trying to build AI as part of a larger community, and we want to share our plans and capabilities along the way. We're also working to solidify our organisation's governance structure and will share our thoughts on that later this year."

OpenAI says the robot will be developed using a current off-the-shelf robot and refined to perform certain common tasks around the house. However, it does not specify the type of robot it plans to create or the tasks it could potentially perform.

The firm also plans to create an intelligent agent with natural language understanding built with the "ability to carry a conversation, the ability to fully understand a document, and the ability to follow complex instructions in natural language".

Inspired by the success of Google DeepMind's AI programme AlphaGo, which recently beat world champion Lee Se-dol in the ancient East Asian game Go, another one of OpenAI's goals is to develop an intelligent agent that can solve and win multiple games.

"Games are virtual mini-worlds that are very diverse, and learning to play games quickly and well will require significant advances in generative models and reinforcement learning," the blog post reads. "We're just getting started on these projects, and the details may change as we gain additional data. We also expect to add new projects over time."

The firm also says it is working on a "living metric" to measure its progress and how well an intelligent agent can accomplish a set task in various environments.

Launched in December 2015, Musk's open source OpenAI programme secured a whopping $1bn in funding from several big names in tech industry including Musk, Y Combinator's Sam Altman, PayPal co-founder Peter Thiel, Amazon Web Services, Infosys and former Stripe CTO Greg Brockman and LinkedIn co-founder Reid Hoffman.

In April, the company created a 'gym' allowing developers from around the world to train their own AI systems on Atari games, environments and challenges. To help speed up AI research and make its development process as transparent and open as possible, the company plans to release most of its innovative research for free as well.

However, Musk has previously warned against the potential risks of AI, which he says could be unpredictable and catastrophic if created carelessly and without regulation.