(A version of this article without the links, recently appeared in the University of Nottingham Alumni Newsletter)
Many companies are rushing to use AI technology. But first things first. Where should we start?
There are tasks that AIs do much better than humans… but there are tasks where humans remain light years ahead of AIs. Understanding which tasks suit AI capabilities, and which tasks are best left to humans can make or break your company’s planning for the next decade. Understanding how to create completely new capabilities, busines processes and busines models? That’s where the fun really starts.
Where to use AI in your business
Artificial Intelligence technologies are great at finding useful patterns in large amounts of data. That is what is useful about them. We need AIs because humans have very limited information processing capacity. Humans get ‘information overload’. We cannot hold that much data in our head, and we cannot process data quickly. We achieve infinitely more when we use AIs to look for patterns in massive datasets – these patterns act as cheat sheets for those datasets.
AI-generated patterns organise data for us. This looks like indexes, contents pages and look-up tables – I recently explained more about how AI-generated patterns in this article.
When to use AI-generated patterns in your business – and when not to
If AIs generate patterns – then when should we use these patterns? The opportunity is simple. To use AI technology for revolution, not evolution. Don’t just automate the old ways of doing things – why not dream up something new?
One definition of resilience that keeps on coming up in my research is that resilience is about bouncing back better than you were before. COVID-19 is the ultimate burning platform – the push for change that we cannot ignore.
So first, focus the power of AI technology on areas which must change right now. What do your customers really really need? What are your staff crying out for?
Second, don’t just make a better version of your old ways of doing things. Take this opportunity to check if what you are doing can have even better outcomes – or if it’s actually needed at all.
For example, you could spend time creating an AI to develop a complex, super personalised and fully automated customer complaints system.
Alternatively, you could simply use your existing data to find the pattern which shows which 10% of the problems in your business cause 80% of the complaints. The same pattern recognition process might also identify which of those 80% of complaints affect your best – and future best – customers.
Or why not pan back even more? Look at the big picture. What is the pattern which connects the most damaging problems in your customers’ businesses with the solutions capabilities that AI can offer your business?
AI enables completely new activities and capabilities.
AIs give your firm new capabilities – like AI vision, speech recognition and personalised recommendations. They lighten the load and they free up humans to do something else.
Machine vision systems identify potential cancer cells. This frees up radiologists to focus on what to do next. In our homes, smart speakers use AIs like Alexa, Siri and Google Assistant to help us in our busy lives. The vast amount of data being generated by Internet of Things devices is too much for humans to analyse on their own. So, organisations really need AIs to support their staff.
Your staff are your scarcest, most valuable resource. It seems odd but your staff are actually the bottleneck which holds back growth. I mean, if you could double the talent in your firm, then you could more than double your growth rate. The better you use and support your staff then the more your business will flourish.
I’m often asked: “What can AIs do better than humans?”
To answer that question, it is best to turn it on it’s head. What are the things that AIs CAN’T do?
AIs do not have intuition. They are not, themselves, genuinely creative. They cannot think critically; and they do not have empathy.
The reasons why AIs cannot do these things helps us see where to use them – and it helps your own job security.
The implication of all of this is that AIs join the team. They do not replace the team. It’s best to think of AIs as new team members (very focused team members…). Deciding how to use your staff in the best possible way is the same as deciding which businesses processes should incorporate AIs.
AIs enable completely new business processes
The saying “Quantity has a quality all of its own” is found Marxist theory and is falsely attributed to Stalin. It’s roots are in ancient Greek philosophy – and it’s highly applicable to 21st century AI technologies.
It turns out that as a thing increases in number or speed, then the qualitative effect produced by that thing radically changes.
Machine learning technologies can integrate many ‘experiments’ – and pick winners in real-time. The autocomplete feature in the Google search box suggests search terms instantaneously, as you are typing words in.
AIs might take time to learn patterns, and they might need expensive retraining. But applying these patterns can be astoundingly fast. Processes can be reengineered on a much more granular level; or even ‘on the fly’- as happens with the optimisation of workflows and layout in Amazon’s sorting centres. Which is a game changer – it’s a process which reengineers a process.
The implication is that AIs can move beyond discrete events to create qualitatively different outcomes. Like new products, business processes and services. AIs can optimise multiple dependencies, like stages in business processes and locations in journeys, in real-time.
AIs enables completely new business models
This is what business model disruption is all about. A business model is a recipe for solving customers’ problems with any capabilities that you can access. New capabilities lead to new business models – which include new problems being solved, or new customers being helped.
New bundles of capabilities produce new ways to create value, as well as new types of value to be produced. This is beautifully illustrated by the capability to fit a powerful computer into a mobile phone, which has led to a multitude of new services based on using them when and where you want to.
New AI-based business models hide complexity for customers – sometimes to the extent that completely new things become possible. Spotify achieves this: the service is the ‘music I love, but that I do not need to choose, or even know that it exists’.
Paying a human to learn someone’s musical tastes; and then pick and produce personalised musical performances is not scalable. But Spotify did it – using AIs.
The implication is that AIs can create totally new types of value – not just new ways of producing value.
How to start implementing AI in your business
The starting point for implementing AI is to change your business strategy to use AI in the best way. Start by building an AI capability in your organisation. Implement the processes which produce AI-as-a-service; to create an AI operation. This means getting the right people and skills mix; and a suitable organisational design, culture, and metric system.
This takes time – but focused projects can be done very quickly, especially when you (temporarily) bring in external expertise.
The best AI tech in the world is just waiting to be rented – like any other cloud service.
The next step is to adjust your business model to make use of these AI capabilities, and make sure your company strategy reflects this.
After this? You need to deploy AI in your company. You can use Journey Analysis (which is similar to business process mapping) to make a use case plan for employing AI and digital capabilities. The aim here isn’t merely to sell the project to the CEO; but to finely plan resources needed, stakeholders to engage, deliverables to measure and successes to broadcast when they are met.
In parallel with the process view that a Journey Analysis gives you, think about your organisation’s design, and the projects that suit your organisational level of data. Or what data you can buy-in.
And all this needs a road map and business case to pitch to your Board/CEO – right at the start.
Have a good journey!
Duncan R Shaw - The qualitative data scientist 