POST is the parliamentary office of science and technology. They are part of parliament but were set up to give unbiased advice to government on current science and technology trends. Additionally, they are tasked with identifying future trends in science and technology. I applied for one of their 3 month internships late last year and was asked to write a POST note for the interview. POST notes are brief descriptions of science and technology topics that may be relevant to current discussions on new legislation. I chose to write one on the use of artificial intelligence in various public sectors.

Artificial intelligence (A.I) is rapidly moving science-fiction into reality. Not only could this increase efficiency and productivity in various industries, but there may be effects on society to consider. This POST note determines the current level and applications of A.I., and explores its implications on society.

Background

Since the first cotton spinning machines, developed during the industrial revolution, technology has transformed industries and the jobs that people do.1 Most manufacturing relies heavily on machines, which have replaced people on the production line. As we move into an age of increasingly sophisticated computers, more jobs than ever before may become affected by truly intelligent, learning machines.

35% of current jobs in the UK are at high risk of computerisation over the coming decades.2 Current legislation does not consider some of the economic, ethical and social implications that the introduction of A.I. could have on people’s lives. Examples of these – discussed in this note – include self-driving cars3, automated news articles4 and financial trading algorithms5.

Current A.I.

The co-founder of Intel, George Moore, stated that computing power doubles roughly every two years with advances in hardware development. This pattern has remained fairly constant since this statement was made in 1975.6 Increases in computing power has allowed machines to simulate, or even better, the thinking abilities of humans in niche areas. The fastest computer in the world (China’s Tianhe-2) runs at 33.86 petaflops (a measure of calculations per second), compared to a human brain which is capable of processing an estimated 36.8 petaflops. The unique feature of A.I. in comparison with regular software, is that it has the ability to learn and adapt, rather than simply repeating an instructed task, and this makes it a much more powerful tool than the previous generation of computers. This could be useful in the following sectors:

Financial

Software has been used in the banking industry for decades. Bridgewater – the world’s largest hedge fund manager – is using advances in machine learning to make predictions of the financial market more accurately than a financial analyst can.7 The A.I. system performs the task automatically, quicker and makes fewer mistakes. Human traders may be replaced by these systems and if this process is not regulated it may lead to unforeseen economic consequences.

Transportation

Companies such as Google Inc., Apple Inc. and Tesla Motors Inc. are aiming to develop fully autonomous vehicles within the next decade for public use.3,8,9 These cars will use advanced A.I. to learn about and react to the environment around them. The technology promises to reduce road accidents and deaths, however some ethical implications apply. Firstly, professional drivers may be replaced (taxis, goods transportation, public transport etc). Secondly, there is currently no legislation to outline who is responsible when accidents occur.

Journalism

Increasingly sophisticated algorithms are being written that can automatically interpret data and produce articles written in plain English. Examples include reporting earthquakes10, financial earnings11 and sporting results12. Increased A.I. sophistication will enable more complex subjects to be automatically reported, without human intervention. As these algorithms become more adept at creativity and original writing, it may enrich cultural science and become a powerful tool for the creative industries. However, considerations will have to be made about the potential effect on cultural development and the lack of human input.

Forecast

It may seem as though progress towards A.I has been slow over the past 70 years and there is debate as to whether Moore’s Law will continue to hold true.13 However, if the exponential growth of computing power continues, true A.I. may only be a few decades away and policies that anticipate the potential changes to society may need to be implemented.

Written by Jake Howden

Endnotes

  1. Marsden, Richard (1884).Cotton Spinning: its development, principles and practice.
  2. http://www.oxfordmartin.ox.ac.uk/downloads/academic/The_Future_of_Employment.pdf
  3. http://www.google.com/selfdrivingcar
  4. http://automatedinsights.com/AP_Case_Study/
  5. http://fsroundtable.org/cto-corner-artificial-intelligence-use-in-financial-services/
  6. http://www.eng.auburn.edu/~agrawvd/COURSE/E7770_Spr07/READ/Gordon_Moore_1975_Speech.pdf
  7. http://www.bloomberg.com/news/articles/2015-02-27/bridgewater-is-said-to-start-artificial-intelligence-team
  8. http://www.theguardian.com/technology/2015/aug/14/apple-self-driving-car-project-titan-sooner-than-expected
  9. http://my.teslamotors.com/fr_CH/forum/forums/tesla%E2%80%99s-musk-sees-fully-autonomous-car-ready-5-years
  10. http://www.latimes.com/local/lanow/earthquake-27-quake-strikes-near-westwood-california-rdivor-story.html#ixzz2wEWmfUmX
  11. https://blog.ap.org/announcements/automated-earnings-stories-multiply
  12. http://automatedinsights.com/sports/
  13. http://www.computerworld.com/article/2503574/computer-processors/physicist-says-moore-s-law-is–collapsing-.html