Science and Society Leveraging from Artificial Intelligence

 

When reading the news in the fields of artificial intelligence, I came across the Wired interview with Andrew Ng[i] – ex-Googler, founder of Coursera, Stanford professor and now the chief scientist at Baidu.

Andrew Ng is one of the most respectful people in the field of artificial intelligence research and applications. This is why, it is striking how calm and reasonable he is about the potential dangers that artificial intelligence may bring. He compares the risk of “AI turning evil” with the risk of “overpopulation on Mars”[ii] and considers artificial intelligence as bare progress (in line with what we wrote earlier) and as incremental advancements, not as a sudden leap that will bring humanity to a new level of societal and technological development.

Another great and striking idea pronounced by Andrew Ng is comparing deep learning algorithms serving the base for artificial intelligence technology with building a rocket ship: “You need a huge engine and a lot of fuel. If you have a large engine and a tiny amount of fuel, you won’t make it to orbit. If you have a tiny engine and a ton of fuel, you can’t even lift off. To build a rocket you need a huge engine and a lot of fuel. The analogy to deep learning [one of the key processes in creating artificial intelligence] is that the rocket engine is the deep learning models and the fuel is the huge amounts of data we can feed to these algorithms.”[iii] This idea has been successfully repeated by many big data developers. While nowadays many company consider data as its precious asset, having data and not being able to explore it and, most important, transfer it into knowledge, i.e. transform and interpret the data in order to be able to make business decisions, is not useful. Collecting big amounts of data and not having a powerful engine for its analysis can be seen as a waste of energy.

We have already covered some details of deep learning algorithms development and applications. The Economist has recently provided more samples of efficient deep learning (including unsupervised learning, machine finding the patterns and applying it to more data corpora) algorithms use: spectacular examples of Facebook’s DeepFace face recognition, Andrej Karpathy and Li Fei-Fei’s (Stanford University) computer vision, technologies used for image recognition behind Microsoft’s HoloLens, machine translation and speech recognition smartphone applications developed by Baidu[iv]. One of the most striking examples is the use of deep learning for fundamental science, where it has a great added value. Thus, CERN, the world’s biggest and most notorious particle-physics laboratory, announced in 2014 Higgs Boson Machine Learning Challenge in order to “explore the potential of advanced machine learning methods to improve the discovery significance of the experiment”[v]. Machine learning has won the challenge – teams of computer scientists had managed to develop statistic tools for pattern recognition and spotting the signatures of subatomic particles without having any particular knowledge of particle physics[vi].

Still, all the scientists and researchers in the field agree that we are still at the beginning of the road – artificial intelligence has many more steps to take before being a replacement for human intelligence. Even the most powerful algorithms are extremely efficient only when performing narrowly defined tasks. Thus, even some of the jobs can be now done by computers the job market is only redefined and re-shaped, not totally taken up by the machines.

Artificial intelligence development is just a new wave of progress rendering our society into the digital economy age where data is crucial for creating knowledge and its exploitation can be automated for a higher efficiency. It also requires developing societal framework, but it is arriving steadily and progressively.

 

 


 

[i] Andrew Ng: Why ‘Deep Learning’ Is a Mandate for Humans, Not Just Machines, WIRED, May 2015, online http://www.wired.com/2015/05/andrew-ng-deep-learning-mandate-humans-not-just-machines/, accessed on May 9, 2015

[ii] Andrew Ng: Why ‘Deep Learning’ Is a Mandate for Humans, Not Just Machines, WIRED, May 2015, online http://www.wired.com/2015/05/andrew-ng-deep-learning-mandate-humans-not-just-machines/, accessed on May 9, 2015

[iii] Andrew Ng: Why ‘Deep Learning’ Is a Mandate for Humans, Not Just Machines, WIRED, May 2015, online http://www.wired.com/2015/05/andrew-ng-deep-learning-mandate-humans-not-just-machines/, accessed on May 9, 2015

[iv] Rise of the machines, The Economist, May 9, 2015, online http://www.economist.com/news/briefing/21650526-artificial-intelligence-scares-peopleexcessively-so-rise-machines, accessed on May 9, 2015

[v] Higgs Boson Machine Learning Challenge, Kaggle, May 12, 2014, online https://www.kaggle.com/c/higgs-boson/, accessed on May 12, 2015

[vi] Machine Learning Wins the Higgs Challenge, ATLAS Experiment News, November 20, 2014, online http://atlas.ch/news/2014/machine-learning-wins-the-higgs-challenge.html, accessed on May 12, 2015