The future of electric sheep

Channel 4’s ‘Humans’ is the latest show to tackle the subject of Artificial Intelligence (AI) and the ethical implications of its creation (Channel4). The premise is simple; in a near-future world, robots or ‘synths’, made to look and act human undertake all the dirty work of the world’s population. The series ultimately explores the moment when the mimicry of human intelligence becomes human, and what constitutes real consciousness, topics that have been explored in numerous sci-fi television shows and films. With the increasing power of computers, this scenario feels exceedingly closer. Now more than ever it’s important to take a look at the possible implications of artificial intelligence, both on us and the intelligence we create.

 Mia, a Synthetic from ‘Humans’

Despite popular belief, Moore’s law states that the number of transistors (tiny switches that are triggered by electric signals) on a microchip doubles every two years, not that a machine’s computing power (Moore’s Law). However, we still observe that computing power is progressing at a tremendous rate, as a new understanding of the nano-scale physical limits of processing hardware is realised. At the current exponential rate of computational advancement, it is predicted AI could reach human capabilities as early as 2040.

But what even is AI? Put simply, artificial intelligences are problem solving devices. There are two types of artificial intelligence, ‘strong’ and ‘weak’ AI (Armstrong). One of the prime examples of ‘weak’ AI is IBM’s Deep Blue (Anthony, 2014), a system that is a master chess player, but certainly does not play in the same way that humans do. Where weak AI aims in getting a system to replicate a desired action such as voice recognition, with the actual process being unimportant, ‘strong’ AI aims at simulating a wide variety of functions that humans can perform.  This is based on the ideology that all aspects of human mentality will eventually be realised by a program running on a suitable computer. However, we have yet to see a real model of strong AI or systems that are actual simulations of human cognition.

The brain: a complex machine

The human brain is an extraordinary organ, the most advanced and complex supercomputer in the natural world, with an estimated operating speed of around 20 petaflops (Westbury, 2014). The cerebral cortex has about 0.15 quadrillion synapses, about a trillion synapses per cubic centimetre. It sends signals to your muscles to move your limbs, decodes images received by your eyes and feels emotion, all without you deciding to do these things. You, as a sentient brain can even think about thinking, and philosophers such as Descartes thought that this self-reflective capability is one of the things that differentiates us from mere animals, that and the ability to use language, coherently forming our thoughts and opinions in order to communicate. Computers can do this too, or at least, can be programmed to appear like they do. In 2014, the Turing test was said to have been passed by a computer program with the persona of a 13-year-old Ukrainian boy, Eugene Goostman, at an event organised by the University of Reading (BBCNews, 2014). The test investigates whether people can detect if they are talking to machines or humans. However, many criticised the test for not being rigorous enough as it measures just one aspect of intelligence. A single test for conversation neglects the vast number of tasks that humans perform easily including vision, common-sense reasoning, or even physical manipulation and locomotion.

Eugene Goostman, Turing Test winner

The Chatbot Eugene Goostman

Another thing that makes the human brain currently function differently to a computer is its ability to parallel compute, that is, the ability to solve a problem simultaneously using of two or more parts of itself in combination. This makes the process a lot faster, similar to employing two builders to build your house instead of just one. The human brain also uses parallel computing in the process of autoassociative memory. You’ve most likely experienced this, to not remember the name of something, only to have it pop into your head when thinking of something else. Part of your brain continues to work on remembering in the background of your subconscious. Platform reported in 2013 on a project called SpiNNaker to utilise multiple processors to compute simultaneously. However, the project is far from replicating the complex web of connecting neurons in our brains.

What happens if AI is achieved?

If AI is ever achieved then we must consider to what extent the intelligence they will be treated as if they were human. This includes whether they should have the same basic human rights as us, calling to discussion the ethical implications of slave labour. Also we will have to consider whether complex relationships could develop between synthetic intelligences and humans and, should that happen, whether they should be allowed to.

Statistics of a Replicant from ‘Blade runner’, based on the book ‘Do androids dream of electric sheep’

Stricter rules must also be outlined regarding at what point a robot is considered sentient. Is it when it passes the Turing test? This has already been surpassed. When it can hold a meaningful, extended conversation with another human, expressing and understanding emotion, humour and self-correction? When it can perform certain tasks as fast as a human can? The best test is a combination of all of these. Many humans struggle with communication with other people, so only testing based on this aspect won’t produce genuinely intelligent results. All these questions of course beg another question: are we just complex machines ourselves? Is our consciousness just a product of extremely complex computation, or is there something that sets us apart from machines? The answer lies in how we categorise our own existence. If we refer to someone, say a murderer, as “not human” we are claiming that they do not deserve such respect. In other words, being human is fundamentally a moral category not a biometric one. This is the best Turing test we have. Should we be trying to mimic human intelligence at all, or would an alternate form of AI be better? These systems could be the best chance we have of creating AI. Although it is unlikely we’ll be faced with these dilemmas soon, or worshipping our robot overlord Skynet, they’re important to bear in mind. No doubt television shows and films will continue encouraging us to do just that.

References

4, Channel 4. Retrieved from Persona synthetics: http://www.personasynthetics.com/

Anthony, S. (2014, December 30). A new (computer) chess champion is crowned, and the continued demise of human Grandmasters. Retrieved from Extreme Tech.

Armstrong, A. (n.d.). Artificial Intelligence – Strong and Weak. Retrieved from I Programmer: http://www.i-programmer.info/babbages-bag/297-artificial-intelligence.html

BBC News. (2014, June 9). Computer AI passes Turing test in ‘world first’. Retrieved from BBC News: http://www.bbc.co.uk/news/technology-27762088

Moore’s Law. Retrieved from http://www.mooreslaw.org/

Westbury, C. F. (2014, June 26). On the Processing Speed of the Human Brain. Retrieved from Blogger: http://chrisfwestbury.blogspot.co.uk/2014/06/on-processing-speed-of-human-brain.html

 

Leave a Reply

Your email address will not be published. Required fields are marked *