One of our partners, Combined Intelligence, were kind enough to write this blog about how to prioritise where to use AI. At Edge Tech we have conversations on a daily basis about AI with our clients, not just about hiring the right people who bring these programmes of work to life but also about where they could use AI. We give them an insight into what our other clients are doing but we also suggest they speak with Combined Intelligence, they have the skills and resources to help you move in the right direction from a hands-on perspective ensuring things are done right the first time around!
Combined Intelligence often get asked where to use AI. This is both a very easy but also a very difficult question to answer.
The simple answer to the question is that you can use AI across the organisation, in all departments, front-office and back-office, from customer engagement to technology infrastructure monitoring. But this answer is not helpful to clients.
But the question is really all about how you prioritise the areas within your business to use AI, and this is where the potential answer becomes rather complicated. There are many factors that will influence the response, from where the organisation is on its AI adoption journey, to its strategy around automation, analytics and business process change around decision making.
If an organisation is asking the question about prioritisation AI use-cases, it is likely that some form of initial innovation activities have been already carried out, producing some proof-of-concept successes that have got the organisation excited about the potential of applying AI wider across the organisation.
The specific approach must be tailored to the details of the organisation structure and culture, but in general, the organisation needs to implement an innovation funnel and prioritisation framework that aligns with its business strategy priorities. The framework will select metrics that are key drivers to the business strategy and overall success of the organisation.
Having an innovation pipeline with a funnel and prioritisation framework will help with the overall process, but there are still many other challenges that need to be addressed.
One of these challenges is that a prioritised use-case may fail, and fail for many different reasons, few to do with the fact that the use-case is impossible for AI to help with. It is more likely to be related to the data available to the algorithm or the quality of the data making the performance unacceptable. Some organisations are culturally unaccepting of failure, and so the concept of a failure creating useful information is difficult to accept. But the reality is these results are insightful and mean more work needs to be done before a successful AI pilot for that specific use-case can be completed. But this is where the prioritisation list comes into play.
Taking a fail fast, learn and move on approach, allows you to benefit from this innovation framework, and move attention to the next use-case in the list, while other resolve the data issues impacting the previous use-case.
For large organisations, multiple innovation teams can investigate multiple use-cases in parallel allowing a constant stream of successful use-cases to be delivered all the way to production environments and benefiting the business.
Many organisations take a controlled approach to implement internal back-office applications first, to help learn the ropes of machine learning with limited risk of impacting client-facing applications and data. This is a reasonable and sensible way to get started but may only give an organisation a minimal level of experience before working on AI applications within the front-office and client-facing. While implementing AI and Machine Learning in a controlled and incremental way may help establish an initial foothold, it will not help with the challenges of scaling up AI delivery across the organisation.
The way the organisation will prioritise AI use-cases will depend on its own capabilities to scale the delivery and its confidence to produce externally facing functionality without affecting its reputation.
There may be other factors that the organisation doesn’t even have the internal resource and knowledge to deal with. Such as correct governance, decision transparency, bias and ethics, that will also limit the speed and scale of AI deployments.
We have for this article ignored the return on investment aspects in the context of prioritisation, this will be covered in more details in another article soon.
At Combined Intelligence we have experience of setting up and running Innovation Centres that have a framework tailored for AI experimentation that delivers repeatable success into production at scale. So rather than asking how do you prioritize where to use AI, the better question to ask yourself is do I have an innovation framework that encourages AI-related use-cases to be explored in a way that allows successful AI delivery to flourish across the organisation in a standard and consistently successful way.