Artificial Intelligence
Artificial intelligence can now create new art we have never seen before and compose new music we have never heard before. Christian Bauckhage talked about the current state of AI at the recent CeBIT conference and the direction of where it is heading.
Christian Bauckhage discussed the current state of AI at the recent CeBIT conference and the direction of where it is heading towards. It talks about all the decision makings based on information could be replaced by AI. Check out the video below!
Bauckhage started his talk with a quote from Albert Allen Bartlett, the late physicist who argued against population growth for most of his career, based on his understanding of exponential mathematics, "The greatest shortcoming of the human race is our inability to understand the exponential function."To regular readers, this statement will seem familiar, as it has been repeated countless times in various incarnations by the likes of Ray Kurzweil, Peter Diamandis and others associated with the Singularity.
For Bauckhage, he unpacks the history of human civilization and relates how technology is on the exponential curve as well, and the consequences are as a result, astounding. "The nature of exponential growth is so dramatically fast, that at the end, they outgrow everything that has happened up until then," he explains.
Related articles
Bauckhage goes on to explain how the technologies of big data, computation and deep learning systems are compounding the exponential growth of artificial intelligence. "Indeed, we are living in very exciting times," he states. "What has happened in the last couple of years is nothing short [of] miraculous.""This future is going on right now, and it's not just three years down the road."
Bauckhage does a good job of explaining why deep learning is causing such a stir and has attracted such interest and investment. "We have seen dramatic progress in the last five to seven years in the area of cognitive computing," he shows. "We have seen breakthroughs in the the areas of automatic text analysis and understanding; we have seen breakthroughs in the the areas of automatic image understanding, speech recognition and robotics."One of the key takeaways from Bauckhage is that even though he is deeply entrenched in the world of artificial intelligence the progress is even surprising him. He teaches game AI, and said as recently as two years ago that artificial intelligence would never beat the top human players of Go in his lifetime. Lee Sedol, the Go world champion was, as was well publicized, trounced by Google DeepMind's AlphaGo system last year.
He also shows another case where researchers at the University of Cincinnati used an AI flight combat simulation system based in fuzzy logic and genetic programming on a Raspberry Pi computer that can beat any fighter pilot. Bauckhage is surprised this story has not had wider coverage.
AI systems are already progamming their own code, and creating art and music too, Bauckhage demonstrates. With open source systems like Google's TensorFlow now reaching wide distribution, the power to create AI systems is set to explode even further. Microsoft and Facebook also have open sourced neural network libraries.
The next big thing, according to Bauckhage is quantum computing. "Artificial intelligence and machine learning depend to a large extent on things we call optimization and search, and this is something quantum computers can do very very fast, so things will continue to accelerate," he predicts.
Bauckhage concludes the talk with a very bold prediction that, "in 10 years from now, everything that is a process—and I think you know that composing new music is a process; doing and accountant sheet is a process; planning a logistic chain is a process—will be done by an AI ten years from now."
He also states that largest artificial neural network on the planet last year had 160 billion synapses, which would be about equivalent to the brain of a bee. Continued exponential growth of neural networks would therefore yield the development of a human-level artificial neural network by the year 2029.
Right on schedule.
Bauckhage is a professor for media information at the Bonn-Aachen International Center for Information Technology (B-IT) and represents the area Multimedia Pattern Recognition at Fraunhofer IAIS. He earned his diploma in computer science in 1998 after studying in Bielefeld and Grenoble; In 2002, he promoted to Dr.-Ing. In Bielefeld, where he subsequently worked as a technical coordinator for a successful EU research project in the field of cognitive vision. From 2004 to 2008, he did research at the Center for Vision Research at York University in Toronto and at the Deutsche Telekom Laboratories in Berlin. His research focuses on efficient pattern recognition and machine learning algorithms and applications of his work can be found in fields as diverse as physics, biology, social media, or business data analysis.
0 comments:
Post a Comment