The small startup plans to revolutionize the field of artificial intelligence with its pioneered technology.
Machine Learning AI that Rivals Google
With a team of only about 70 people, Sentient has managed to link up hundreds of thousands of computers to function as a supercomputer that surpasses that of Google. Founded in 2007, the private company has received over $140 million in funding, much of that being received in the past year.
Sentient focuses on running a method of machine-learning called evolutionary algorithms. It uses its large base of computers to create an array of different algorithms for a problem, and then uses the best results as starting points for the next iteration of algorithms over successive generations. In the past, Sentient has done research to predict the development of sepsis in intensive care patients, but they have since moved on to larger and more ambitious projects.
Image Credit: Shutterstock
Sentient Connects One Million Idle Computers
While Sentient is relatively private about their infrastructure and current work, they explained how they managed to get hold of their extensive computer base, which can have up to one million processors working on a problem for months at a time. Led by Adam Beberg, who pioneered the concept of linking computers to unanimously work on a problem, the company states that many of the machines are idle computers inside data centers, the facilities that power websites and mobile apps. The centers have large amounts of unused machines since they are only used to handle surges in public usage, such as on Black Friday. Sentient can then connect the machines over the Internet to run machine-learning software as a single supercomputer.
Sentient has earned revenue from running financial-trading algorithms, but is moving on to offer services to the healthcare and online commerce industries. The company has not released information on the software it is currently developing, but it is designed to keep its data highly encrypted for private clients.
The Future of Artificial Intelligence
Sentient is now trying to implement its cross-computer linking process to deep learning, which is the main subject of most contemporary AI work. However, the company will likely have to rethink their programming to achieve this. Sentient excels at tasks that can be split into individual pieces that computers can do without communication. Deep learning, on the other hand, requires constant flow of data between the processors.
The company has already managed to have thousands of its processors working together with deep learning. A Google spokesman commented on Sentient’s efforts stating that the results of deep learning are more significant than the power of the computer system behind it. Still, imagining what Sentient will be able to do with a network of hundreds of thousands of computers working at once is very impressive. Deep learning has produced huge advances in image and speech recognition, and though the company is extremely private with its current projects we will inevitably see rapid growth in distributed AI. This is certainly a company to watch!