AI hype is at fever pitch. It’s slated to fundamentally change everything about our world, from our economies to the way we get around cities. But how much of the hype is credible, and how much will AI change the nature of business in the near future? Will AI completely take over the realm of human expertise?
Algorithms are changing the world. No doubt. One day, we’ll consider algorithms as tools as transformational, significant and epoch-changing as the wheel. Not only do they make basic computing possible, they’re bringing the reality of sophisticated artificial intelligence closer every day. No wonder PwC says up to 30 (opens in new tab) per cent of jobs in the UK will be replaced, either by basic automation or artificial intelligence. However, unchecked praise and hype around AI can blur our thinking about the foreseeable impact of these technologies on our advanced industries. We can be lured into saying things like “AI will change every job fundamentally”. Well, in advanced industries, humans are still vital, and will be for the future. In other words, the ballooning hype around AI should be popped.
AI’s advances have 3 significant drawbacks for business
Problem 1: It just “feels wrong”
Just think back to a few years ago when you typed in a message into Google Translate. No matter what two language combinations you selected, the input and output terms didn’t exactly match. Translating from my native German into English, for example, produced interesting results, including an inability to translate idioms. The leap in Natural Language Programming has been in part due to a change in its structure, from translating a single sentence in one language into another, to using Neural Machine Translation, which more accurately translates meanings. It does this by using machine learning via neural networks. Indeed, this new enhanced learning capacity has made the output of translation services a lot more convincing (and idioms are now also covered). But often, the tech still fails the vital test: we can “tell” if the text was written by a machine. This can be partly explained by the machine’s inability to capture higher levels of self-reference. Furthermore, ROBO-calls and chat bots (especially when customers want to make a complaint) are also sources of frustration: especially since they simply don’t recognise or fulfil the customer’s emotional need for a quick response.
Problem 2: Lack of data
There are many other examples of success in AI and machine learning, particularly when it comes to self-driving cars. But some fear an “AI winter (opens in new tab)” is coming: that is a stalling of the learning potential of the machines, due to the lack of data. Machine learning has advanced in the area of translation, since there is so much data to utilise, and a failed output is simply a bad translation. In the case of self-driving cars, there’s simply not enough data so far for the algorithm to learn and generalise situations. For example, in one unfortunate example, a self-driving car plowed into a white truck, interpreting the colour of the truck as “sky” and powering ahead; the human driver of the self-driving vehicle, Joshua Brown, unfortunately lost his life. One of the questions researchers have had to face is if the algorithm must repeat this accident thousands of times before it learns to differentiate “bright sky” from “vehicle-white”.
Problem 3: It costs too much
Finally, there’s the case of Go. Go is an Asian strategy game and has been called the most complex game ever designed. Not bad for a game consisting of a wooden board and white and black counters. A documentary was made about the success of a team building an AI which beat the world’s second best ranked Go player, Lee Sedol. It was heralded as an unmitigated triumph for AI. However, once you factor in the monumental costs and energy it took to accomplish this task, it becomes clear that this is simply not yet a business reality for us.
AI is not the future of work for the foreseeable future
Each of these examples has been heralded as an advance for AI; yet each demonstrates failings which would be disastrous for business positions requiring expertise. Firstly, leadership and expertise should make us feel comfortable; we should not feel a lack of trust, as we do with translations. Secondly, we should be able to handle business-challenges spontaneously, and not have to wait for a fail-safed solution before we can offer advice. Thirdly, business expertise should be affordable, and a business reality. Since none of these things is covered by AI thus far, it’s pretty safe to assume humans will continue to be necessary for jobs which require a certain degree of business expertise and experience. Though it will catch-up one day, and no one should feel too comfortable in their positions, right now the reality does not match the hype.
Working with and not being replaced by AI
Instead of focussing on how AI will replace every job, we should think of how we manage resources, and better enable human expertise to flourish in work. AI can do jobs for us which require repetition, leaving us free to tackle more creative parts of our work. We should, in other words, think of how we can work with AI.
Technology and AI could help structure the working pattern and communication of humans who are in expertise positions. All jobs, no matter how advanced, are made up of functional elements which can be structured. For example, organising people, scheduling and communicating tasks to a team (as opposed to handling complaints) can all be structured by chatbots and other tools, leaving more creative tasks to the human. It may also be possible for more sophisticated, but still repetitious, parts of a job to be structured; a promising example from the medical profession demonstrates the ability for mundane tasks such as reading X-rays to be carried out by machine-learning software, leaving the doctor free to carry out more difficult tasks (opens in new tab). She could attend to patients, for example, while the AI is left to determine whether an injury is a break or a fracture.
What’s more, technology can be properly deployed to source and order expertise where it is most needed. For example, Flash Organisations, the Stanford University-researched (opens in new tab) method to effectively solve business problems. Flash organisations structure expertise in a hierarchy to solve projects, and Stanford argues that these teams are only effective today because we have the technology to effectively source the right experts for positions. We could further enhance such teams by using a mix of AI and humans in the team: the humans could define concrete goals for the AI, and it could collect and annotate the right data for it, making problem solving a lot easier.
In the pursuit of successful learning based technologies, we should also retain perspective, and rethink how we use the resources we currently have. While there’s no doubt AI will change our current ideas and structures in unforeseeable ways, the fact remains that humans will be vital for business at all levels, but particularly for jobs requiring expertise.
Christoph Hardt, co-founder and managing director, COMATCH (opens in new tab)
Image Credit: Shutterstock/Mopic