Elon Musk keeps surprising the world with his technological breakthroughs. OpenAI, a non-profit company focused on AI research, recently made an announcement regarding its groundbreaking AI invention. It has developed an AI system that can complete an actual physical task after watching just one demonstration of the task.
Inside the OpenAI system
At the core of the AI system are two different neural networks — a vision network and an imitation network. These two work behind the scenes to provide the remarkable capability to imitate human actions, a giant step closer to building true AI systems. A robotic arm repeats the process of picking up blocks and stacking them in a particular configuration. It does this by witnessing just once a simulated demonstration performed by a human using a VR headset.
Researchers have used thousands of simulated images to train the vision network. These images represent various arrangements of objects, the ambient lighting, colors and textures, among other factors. The robot’s camera captures the objects’ positions relative to it. It then uses this information to pick up the objects. As the images are not real but simulated, a vast number of images can be simulated and used to train the vision network.
An imitation network, the second part of the puzzle, is responsible for mimicking the demonstrated tasks. Similar to the vision network, thousands of virtual simulations are fed to the imitation network to train it. Once the imitation network gets trained, it can easily replicate similar demonstrations by just watching them once.
An important point regarding this unique achievement is its broad applicability. Although the demonstration used blocks, the learning behavior and the AI capabilities are not specific to blocks. The system is applicable to a large number of functional areas.
Recent OpenAI offerings
One of the most influential visionary of our times, Elon Musk is the co-founder of OpenAI. Unlike Musk’s other business, however, OpenAI is non-profit and aims to ensure AI benefits have a wide reach.
Its other achievements include Universe, an open source platform, using which developers can train their systems to accomplish computer-based tasks the way humans do. In addition, Roboschool, used for robot simulation, offers several environments for users to test and train their robots. Education is another focus area for OpenAI.
AI, in addition to IoT, is one of the most exciting future technologies that will play a key role in shaping the future. A world where systems are capable of teaching and learning from other systems will unleash infinitely exciting opportunities.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.