Artificial Intelligence (AI) is simply intelligence in machines. This contrasts with natural intelligence found in humans or other biological organisms. In 1956, artificial intelligence was established as a field of formal research. Initial work resulted in new tools that could solve mathematical problems. Researchers discovered that AI creation was challenging, and progress slowed in the 1970s. Recent advances in computing power and the availability of large data sets have opened up new avenues for AI.

Notably, scientists have made considerable strides in one specific application of AI, Machine Learning. Machine learning is a system that learns from the data it analyzes and then uses the results to improve future work with the data. Machine learning estimates the trip’s duration and provides route guidance. Google Maps uses machine learning to analyze changing traffic conditions and incorporate user-reported information to help reduce travel time and recommend fuel-efficient routes. These well-defined, specific tasks can be broken down into logical steps. Machine learning works best.

Supercomputers such as the Department of Energy’s Aurora at Argonne Leadership Computing Facility will make artificial intelligence more powerful and efficient. Credit: Image courtesy Argonne Leadership Computing Facility

Deep learning is the next level in AI applications. Deep learning is a complex task with many variables that are difficult to pin down. It uses a neural network to create computers that function like our brains. Deep learning uses many neural networks stacked on top of each other to build a complex network. Deep learning requires a lot of computing power to train models from large amounts of data and then use those models to make decisions. Digital assistants such as Amazon Alexa, Apple Siri, and Google Assistant are a few examples. These assistants can listen to what you say and determine what task or information you want.

Artificial Intelligence Facts

We can see AI’s ability to play games as one way we can understand its progress.

• 2011: Jeopardy! IBM’s Watson won the U.S. game show
• The Google DeepMind AlphaGo AI defeated a human grandmaster in the Chinese complex game of Go.
• Humans say AI doesn’t excel at games that require teamwork.

DOE Office of Science: Contributions towards Artificial Intelligence

Since the 1960s, the Department of Energy Office of Science (DOE SC) has been involved in research on AI. ASCR developed several technologies–including massively parallel input/output systems and linear algebra routines–that led to today’s AI systems. ASCR’s recent role in high-performance computing and Exascale Computing will assist in building the hardware and software required to power future generations of AI. Other DOE SC programs also benefit from the ASCR program’s support. For example, the DOE SC Fusion Energy Science program uses AI to control fusion reactions to make fusion electricity production a reality.