Let’s be straight up front: there is simply no intelligence in artificial intelligence (AI). This is because today’s AI is made up of very sophisticated algorithms, achieving the outcomes through heavy calculations. The result: it is great at handling routine, regularised, labour-intensive tasks with clear goals but lousy at taking over any open-ended and non-standardised processes.
There is simply no intelligence in artificial intelligence.
The true source of value
AI alone is therefore no guarantee to creating value for organisations. Instead, the true power of AI lies in its use in collaboration with humans. We will always need humans to translate and expand the latest technology. And we will always need humans to make meaning and connections that inspire us and connect us the moment we put out technology down. Microchips and robots cannot and will not replace relationships.
This is why the best way to realise the potentials of AI is “human-in-the-loop”. Granted, for many activities, it makes complete sense to use algorithms to fully automate the repetitive work done by humans (e.g. staff to “eye-ball” the details on written documents and manually key them on the computers). Yet, to take full advantage of AI, it is necessary to involve humans to take on those tasks that machines cannot easily do (e.g. automating the standardised processes but using humans to control the quality of output, which in turn improves the algorithms’ performance). Often, artificial intelligence is only good when human intelligence is also involved.
Power to the people
Many people believe that the more powerful the AI technology, the more they will be able to achieve. This is wrong thinking: it is not the raw power of the AI that matters. What is far more important is to think through the role of humans in the setup surrounding the AI, the underlying workflows and processes as well as the supporting IT infrastructure. Put differently, an arrangement that machines and people work well together will champion over the use of the most powerful AI with no human supports.
The way to go forward is not so much about artificial intelligence as collaborative intelligence.
Blackbox vs. Whitebox
There is another important benefit of getting humans and machines to collaborate. Typically, in machine learning, we can only see the inputs and outputs of algorithms. How they get to the observed results is frequently a complete mystery to us. Applying such “blackbox” algorithms in various aspects of our daily lives, such as in the justice system, can lead to huge social and ethical issues. Neither is it acceptable for businesses: can managers justify using algorithms without being able to clearly and fully explain how they work? Ignoring “blackbox AI” would only bring more costs, operational inefficiencies, reputational damage, and even legal repercussions. A much better alternative is “whitebox AI” – building a system that is understandable, justifiable, and auditable by us. In many ways, we are much better off having an AI-driven setup in which many simple activities are done by AI, leaving the rest to humans as opposed to having the hugely sophisticated algorithms to run the entire setup.
A tool by itself is useless. It takes a person to make the most out of it. AI as it stands today is nothing more than a tool. Growing, changing, exploring, boundaries pushing can only be achieved with a symbiotic relationship between humans and machines. The way to go forward is not so much about artificial intelligence as collaborative intelligence.