Following on from my last blog – Why Emerging Tech is Failing to Deliver Expected Returns, I wanted to provide some balance by providing a few suggestions to how to avoid failure when implementing cognitive technology. In a recent McKinsey article, they shared the normal way cognitive technologies are introduced to a business, and why this approach fails. Largely the reasons come down to six common elements:
- Senior management are not educated on cognitive technologies, and their likely impact. It’s important that an organization understands the proper uses of each technology and the best way to employ it. They also need to understand the skills required to engage with the technology – are data scientists needed, or is the user interface sufficiently human friendly to allow competent analysts to perform the desired functions?
- Selecting the wrong type of technology for the business – which could be rule-based expert systems, machine learning, deep learning, natural language processing, or robotic process automation. In some cases, multiple types may need to be combined.
- Learning is not co-ordinated across the enterprise – identifying expert skills in the organisation and forming some form of centralised organisation to ensure that learning is shared across the business. Ensure senior executives get regular briefings on advances in cognitive technology.
- Going for the big show – attempting to wow the management team often ends up slowing progress and losing confidence. Instead, focus on a small scope that automates a single digital task, where you can readily combine human and machine-based expertise. This is more likely to gain support for the machine, and dispel fear of human job replacement. Leverage this success quickly across a portfolio of small pilots of proof of concepts. Then scale up those that work, and ditch those that don’t.
- Not readifying the business – the goal of cognitive technologies is to enhance human performance, provide insight, or to automate some activity. There are strategic, process and data elements that need to be considered ahead of any implementation, regardless of type. Attempting to implement a cognitive technology without due consideration to strategic alignment, process optimisation or data preparation is setting oneself up for failure. Even if you feel you are not yet in the position to invest in cognitive technology, there is a lot that can be done to readify the business for it. This then makes any AI implementation as easy as possible. Remember – you only get one chance to make a good first impression.Discounting the importance of IT in the end to end selection and proof process – in many cases, cognitive technology requires integration with existing systems and other types of technology. Sometimes, such integration can be complex. IT is your best friend – involve them early on and provide sufficient time for them to perform the necessary work.
- Not allowing time for tuning the machines – cognitive technologies learn, adapt and improve over time. Set expectations that roll-out is purely a project milestone, not an optimised performance point.
Of all of these, I believe that not readifying the business for AI is the biggest, and most costly mistake. Cognitive technology is only just at the beginning of its operational evolution in most businesses. This is a time of trial, where we can learn, make mistakes and even fail. But, if you are going to fail – fail small, and fail fast. Preparing your business for AI is the best way to ensure success.
Read McKinsey’s Beyond “doing something cognitive”