Comment

Wider implications of AI

Automation is happening. We can deny it no longer. This year, the world’s largest electronics company, Foxconn – known for manufacturing Apple, Google and Microsoft products – laid off 60,000 factory workers to have them replaced by robots. Others will follow suit.

The Queen announced the government’s plans for new legislation making provisions for driverless vehicles on the roads of the UK in the State Opening of Parliament. Such legislation will doubtless lead to mass layoffs in the transport industry, currently employing 4.6% of the UK population. Once the cost of technology drops below the cost of a wage, redundancies become inevitable. As a result economic models such as Universal Basic Income are now being considered by some in a more favourable light, though not by nearly enough.

Automation is the most imminent threat to work, both in the UK and abroad, but another threat looms on the horizon: artificial intelligence (AI). Yes, manual labour jobs will soon be lost, and this could not be more devastating given the government’s shameless selling of public assets, cuts to welfare, and dissuasion from higher education, but AI threatens white collar jobs and creative jobs too. And that is only the start.

This year, Google DeepMind’s AlphaGo program beat Go champion Lee Sedol. DeepMind are also working in conjunction with the NHS on Health, a program which uses patients’ blood samples to predict the likelihood of kidney injury. The project aims to diagnose illnesses with higher reliability than doctors. In addition, IBM’s Watson has been suggesting treatments for individual lung cancer patients since 2013.

Less well known are computers’ recent endeavours in poetry. The researchers at Google Brain recently demonstrated the results of a program given a series of romance novels. Tasked with inventing ten sentences of its own, the program used these novels to construct its own understanding of grammar and produced a number of syntactically correct sentences. Amusingly, these anaphoric sentences were decidedly afflicting. One poem ends “I had to do this./ I wanted to kill him./ I started to cry”.

Worth bearing in mind is that growth demands the development of these technologies. They will improve and they will be used frequently when they are good and cheap enough. What is most telling is Google’s investment into them. Even if you think the Chinese Room thought experiment was a good argument for why AI will never happen, when it comes to technology, Google has a reputation for being ahead of the curve. So I’m going with them on this one.

Once programs are able to write fluently, passing the Turing test becomes meaningless. We will not know what was written by a person and what was written by a machine. Humanity has never been more connected than it is now, yet all that connectivity could be undone if we lose trust in written language. If power is a discourse, AI are prime candidates for that power. The rise of the machines will be persuasive rather than violent.

AI may be in its infancy, but the questions it provokes must be answered soon. When humanity discovers a technology which makes things more convenient, it never lets go. We didn’t need agriculture until we developed it, nor writing, cars, smartphones or oil drilling. These technologies all necessitate themselves and AI will be no different.

14/06/2016

About Author

Avatar

matthewbiggs