In Part 1, we discussed why you shouldn’t be worried about automation.
So, how is the present different from the past? Artificial Intelligence!
Artificial Intelligence(AI) is revolutionising every industry and it is happening at a breakneck speed since its beginning 70 years ago. In early 2011, IBM’s AI named Watson, competed and won the Jeopardy! championship. In March 2016, Google DeepMind’s AI, AlphaGo, defeated the Go world champion in four games to one. Experts predict that AI will outperform humans in various tasks such as language translation (by 2024), driving trucks (by 2027), replacing retail workers (by 2031) and performing a surgery (by 2053). As Elon Musk said, “there will be fewer and fewer jobs that robots cannot do better.” According to a 2013 Oxford Martin School study, 47% of the current jobs could be automated by 2034.
There will be fewer and fewer jobs that robots cannot do better.” — Elon Musk
The New Division of Labor published in 2005 by Frank Levy & Richard J. Murnane, states that automated driving would be extremely difficult to achieve due to many circumstances and factors to account for. However, both authors underestimated the effect of Moore’s Law which states “the computation performance would double every eighteen months at the same cost”. Google’s autonomous car, Waymo, took seven years (2009 – 2016) to reach two million miles. In 2017, it logged a million miles in seven months alone, further proving the effect of Moore’s Law.
In Part 1, we mentioned that since the Industrial Revolution, new jobs and industries were spawned due to automation. So how is it different this time? From the above-mentioned examples, AI has evolved from manual repetitive machines to human equivalent systems. Observational Learning is a unique trait that makes the human being unique and enables them to dominate the planet. Machine Learning, on the other hand, allows for AI to learn from data, identify patterns and make decisions, without being explicitly programmed. This allows machines to shift from being a specialised machine to a general-purpose machine. Computing was a big thing in the 1950s. However, it was costly and it only did a specific task. As the Central Processing Unit (CPU) got more efficient, its cost decreased. As it was increasingly designed for general purpose computing, the adoption rate of the personal computers increased exponentially. The same fate could be true for AI machines. As AI continues to progress in artificial general intelligence, AI would eventually be able to function over a large range of domains and dynamic situations, very much like a human.
While a human worker has increasing wages, falls sick and makes human errors, a machine robot could likely have a lower recurring cost and be more operationally efficient. In an idealistic world, the latter could be a more ideal candidate for any job – white-collar jobs such as doctors and programmers may not be spared either. Google’s AutoML system was developed as a solution to the lack of talents in the AI programming. The machine-learning codes made by the AutoML was found to better and more efficient than those made by the researchers who created the AutoML.
Essentially, AI is making better AI than humans!
Since the Industrial Revolution, we worry that AI would replace human workers. Today, humans use machines as tools for better productivity and efficiency. With the exponential growth of AI, the future should not and cannot be compared to the past. Technology evolves faster and cheaper in a way that the human biology evolution cannot match. It’s only a matter of time that AI will soon replace the human mind.