Embracing AI

Practical steps for adoption

In part one of this series we discussed looking at the social and routine nature of work to understand where AI could play a part in boosting productivity.  As a reminder, activities and tasks that tend to be routine in nature are more suitable for either automation (where the tasks are routine and performed in isolation), or augmentation (where the tasks are routine but done together).  Analysis of the nature of work in a real estate business through mapping 300 processes that formed the organisation’s value chain, revealed high-level results that 38% of tasks were routine and done apart and therefore ripe for automation, with a further 17% of tasks being routine and apart, and therefore can be augmented using AI.  This analysis was backed up a McKinsey study where they estimated 40% of work in the real estate sector could be automated using technology that is available today (such as Robotic Process Automation (RPA), and Machine Learning (ML)).  For interest, our analysis showed that 25% of tasks were varied and together and therefore likely only ever undertaken by a human.

Using this framework is a good starting point for considering the deployment of AI tools in your organisation.  Through mapping the nature of work on your current processes, you can efficiently determine the areas most suitable for implementing new tools.  First, focus on deploying traditional AI tools such as RPA and ML to automate tasks that are routine and apart before looking at deploying tools such as Retrieval-Augmented Generation (RAG) for augmentation of tasks that are routine and together.

The risks of adopting AI

There are a number of risks that need to be considered when deploying AI:

Ethics

Trust in a high level of AI sophistication can be dangerous.  There are many stories on the quirks of large language models (LLMs) and hallucinations are common (where a LLM can make up something that isn’t true).  The ‘black box’ nature of a language model makes it difficult to trust the answer it gives and this is particularly important when using a model to interpret customer data.  It is important to determine your ethical position on what types of data should be included in your model and make suitable disclaimers where AI is used to process information that is shared with others.

Regulation

Regulation in the use of AI is being developed and is necessary to provide a framework for businesses.  The EU’s AI Act is perhaps the first comprehensive AI law and has been enacted in 2024.  It contains a list of prohibited uses and also identifies high-risk AI use in areas such as hiring, human resources and labour supervision.  Penalties for misuse are steep and more information can be found here.

The human factor

As discussed in part one, the use of AI will likely bring competition for the jobs of knowledge workers as the gap between highly skilled and average skilled workers narrows.  Low skilled jobs may be easy to automate however the cost of this automation can often be more expensive than employing human workers.  At the other end of the spectrum, high skilled, high experience jobs (together and varied) are unlikely to ever by replaced by AI and this leaves the roles at the middle of the organisation most susceptible to change.  Automation will increase productivity and create the capacity for more value, but it may result in fewer people doing the remaining work, leading to a shift in the workforce distribution (the dumbbell effect).  The skills of those jobs in the middle are likely to be different than today, where people will need to know how to work with AI to be able to be successful.

In addition to the impact on the organisational structure, the act of transforming a business using AI tools requires careful change management.  Both resistance to change and excessive enthusiasm for experimenting with AI represent two extremes, neither of which are likely to result in positive outcomes.  Creating strong AI leadership and being clear about your organisation’s vision for AI is critical for success.  Furthermore, trying to simply overlay AI into the systems, process and data that you already have will likely lead to frustration, failed experimentation and disillusionment in the opportunities that AI can provide.  At Bothy Consulting, we specialise in advising businesses building the capabilities to get these transformation initiatives right.

Conclusion

This series has provided guidance on using AI to boost productivity through turbo-charged knowledge management.  With productivity growth declining over several decades, businesses need to look at deploying AI tools to help reverse this decline.  There are many AI point solutions available on the market that are focused on solving very narrow problems and it is important to take a step back and consider how work is carried out in your organisation in a platform way to target the areas that will reap the benefits of AI.

Bothy Consulting is an innovation and transformation consultancy that provides advice on strategy development, process mapping, organisational design and Innovation-as-a-Service.  Its founding partners have over 50 years of combined corporate experience and have led many transformation and change initiatives and developed digital and AI strategies for leading businesses.  If you would like to hear more about their services, email enquiries@bothyconsulting.com