sHANNON SMITH QUESTIONS ARTIFICIAL INTELLIGENCE
As Amazon’s Alexa makes ‘herself’ increasingly at home in the world of the domestic, the possibility of her and similar forms of artificial intelligence in the workplace is fast becoming a reality. Our dependency on the virtual assistant is nothing new – we have been relying on Apple’s Siri since 2011. AI also has the potential to completely transform our standard of living, to create a shift in human capital away from mundane transactional work. However, this begs the question: what is replaceable in the workplace? What if artificial intelligence can go beyond mundane transactional work, if it could replace, for example, your doctor?
One University of Oxford student* has provided their unique insight into robotics and artificial intelligence in the work place, having worked in the front line of the manufacturing industry before moving into medical AI.
Artificial intelligence has long been a point of fascination within both the world of science and the media, for example the film industry. We have been presented countless apocalyptic visions and utopian dreams. One of those dreams is AI’s ability to cure. Do you think we will see a future where AI replaces, for example, surgeons?
Firstly, there’s AI that specializes in recognizing images and the way that it does that is though something called the deep neural network. The deep neural network will feed millions of images to [the machine]. Say for example, you want to classify toasters; it’s seen more toasters in the world than you probably have because it’s seen everything on Google. The neural network will pick up features of this toaster, for example dials, and it will do it in a much more mathematical and geometric way, it will pick up random abstract things. If you add or change one pixel in the image, you can totally confuse it. There was a video that was released of a network that was classifying an image of a turtle – they changed one pixel in the image and it became a kettle. So it’s obvious what we are doing now does not mimic how the brain works, because if you were seeing a picture of a turtle and one-pixel changes, you still know a turtle when you see it. This is not how the brain works, so the way we are approaching it is incorrect.
So this is why AI can’t be used in surgery right now, because it wouldn’t be able to recognize the human body in all its variations?
There is no way AI can replace a human doctor. For example, if there’s something like filling in a form [in a hospital], that’s a pretty simple task, that it could do. However, AI lacks the instinct that humans have. How do you program the instinct that is necessary for surgery? When a baby is born, how does it know it has to cry? No one has taught it that. It’s hard programmed into our DNA but it’s a form of intelligence we don’t yet understand. We are nowhere near replacing humans with robots in hospitals.
You also have experience in the car manufacturing industry so you are familiar with the capabilities of robots to replace workers. Is there a lot of pressure to reduce the number of human employees in favour of machines?
A big thing in these corporations is that they want to get rid of the ‘heads’, they have an estimate of how much they spend on each worker per year, including pensions and so on. Robots don’t need pensions. If one robot costs £100,000 and can replace two workers then the management will give you a big pat on the back. In some places of work, they will actively financially reward you for cutting down the workforce. However, the way we used to work around this pressure was to add five fake people to a design and then when the management requested cuts in the workforce, we would gradually get rid of these fake people, one every time the managements makes a request. Its entrenched so deeply in the culture, you know before hand that they are going to tell you to get rid of people, even if they are people you need.
* The contributor has asked to remain anonymous.