YEREVAN, Armenia -- The Power of Decentralization: Promise and Peril. This is the theme that kept busy global IT leaders at the 23rd World Congress on Information Technology (WCIT 2019) hosted by Armenia in its capital city, Yerevan.
How information and communications technology (ICT) is transforming our lives, and how the industry is preparing for the radical change that Artificial Intelligence is bringing to all sectors took a predominant role in the discussion.
During his keynote speech at WCIT, Murthy said that "technology has the power to make life more comfortable for human beings, as long as it is put into good use." Speaking about the benefits of adopting autonomous vehicles, Murthy said that 94 percent of accidents are caused by human error. "Autonomous cars will reduce accidents, reducing deaths caused by car accidents."
Rise of the machines: The price of creating power
Technology has always the power to make life more comfortable for human beings as long as it is put into good use. -Narayana Murthy, Founder of Infosys
Big Data, Artificial Intelligence (AI), and Machine Learning (ML) offer the promise of unrevealed insight and efficiency; robotics, the promise of freedom from physically dangerous or taxing manual labor, all in ways never before imaginable.
However, at what price? The widespread deployment of increasingly sophisticated Big Data, AI, and automated robotic systems threatens to make entire categories of workers redundant by automation.
Big Data and AI systems also threaten to distort the human decision-making process, subordinating the role of human judgment.
And paramount questions arise; should the cold logic of hard data be the master of human systems? What room will remain for judgment, morality, and human compassion? How much authority and decision-making are humans willing to cede to machines?
Where and when will it be necessary to draw the ethical and practical line in the application of Big Data and AI in areas such as medicine, where compassion and morality ought to reign over clinical statistics?
How do we avoid being ruled by Big Data, or automated systems? How do we control AI systems, already so complex that no single person can understand them, and keep them from going rogue and turning on us? These are some of the questions everyone involved in the creation of AI and all those concerned about technology going wrong should ponder. The topic was deeply discussed by experts on the subject at WCIT.
AI: What is your fundamental fear?
Richard Quest, Business Anchor for CNN, moderated the panel integrated by James Bridle, Multidisciplinary Artist and Journalist; Martin Ford, Author and Futurist; Daniel Hulme, Director of Business Analytics MSc, University College London and CEO of Satalia; Christopher Markou, Ph.D, Leverhulme Early Career Fellow and Affiliated Lecturer at Jesus College University of Cambridge; and Narayana Murthy, Founder and Chairman Emeritus of Infosys.
Richard Quest asked the members of the panel what their fundamental fears about AI are. The panel established that as AI, Machine Learning, and robotics advance, more jobs will be lost. "That can be any job, including some white-collar jobs," said Martin Ford.
And although more jobs, other different jobs will be created, are those new jobs going to be enough for everybody?, he pondered. And, what about the transition period? What are the big potential challenges that will be occurring in the next decade, or two?
"Companies must make profit and create jobs," said Narayana Murthy. "According to research from Oxford University, Murthy said," 40 percent of jobs will be automated by 2025.
"Regulation is good when it's not telling you what to do," said Christopher Markou. Discussing the limits of these machines, he added that AI should not exist in places such as classrooms. "Where we don't want these things is what we should be discussing," he said.
AI machines are predicted to be the last invention of human beings, and this could happen in our lifetime. "Adaptable machines can be dangerous. If the machine, say autonomous weapons, have the capacity to adapt to their environment and learn from it, then if the machine is in a bad environment learning from humans whose purpose in life is to damage other humans it means that is what the machines will learn. And that can be unstoppable. Indeed."
In the end, Richard Quest ended the discussion by asking the panel if every machine should have an ON/OFF switch. Answers varied. Based on AI safety research conducted by the University of Cambridge, "the central authority must remain human," Christopher Markou concluded.
What do you think, should every machine, including AI machines, have an ON/OFF switch?