Artificial intelligence for commercial application has come a long way: With the power to forge ahead to transform how businesses operate and thus reshape the way humans work, there is no question that the creation of AI and the speed of its own advancement in this rapidly progressing and challenging industry is staggering.

Two watershed moments have put AI on the map in 2016: Its recognition by the World Economic Forum as one of the key trends for the next 20 years, and Google’s DeepMind platform winning a game of Go — a simple game that has been notoriously difficult for computers to master because of the sheer number of potential moves.

We want to welcome you to take a deep dive here and explore how the beauty of technology and the power of AI will unfold, as well as understand why strategy games are often used to benchmark AI. On this website you can read more about the latest AI news digest/research digest and get a better picture through commentary of real AI experts. You can also read more about our own AI research and news about HIRO (Human Intelligence Robotically Optimized)™, Aragos AI-based automation platform in the news/research section.

Understanding Past Drawbacks

In the first golden age of AI (a branch of computer science attempting to build machines capable of intelligent behaviour), expert systems were at the center of attention. While still being an active research field today, their commercial applications have been very limited. Several drawbacks resulted in its decline:

  • Closed-world assumption: Everything has to be defined from the beginning.
  • Incapable to adjust to changes in the environment.
The Importance of Deep Learning

Big technology companies like Google, Nvidia and Facebook are currently working on developing deep learning; allowing computers to learn the way a human would in order to progress what many are calling the next revolution in technology – machines that ‘think’ like humans. However, there are also serious limitations to deep learning

  • Accuracy depends on how well the training-set captures the real-world scenario.
  • Intransparency of the learning process: Acquired skills cannot be extracted, re-used or optimized.

At Arago we are driving a different approach: We start from reasoning with a knowledge based problem solving engine. This allows us to use formalized language to teach knowledge to the machine. We avoid the typical pitfalls of classic expert systems by allowing for incomplete information and ambiguities in possible solutions. Our experts are using machine learning approaches to optimize the solutions by the problem solving engine. And we are using neural networks for perception, i.e., to recognize patterns in large streams of unstructured input data.