top of page
  • Writer's pictureWilliam Webster

AI – Experiment at the Top

Updated: Jan 11


AI is coming for knowledge work, the problem is that we don't know how it will affect us and the businesses we work in. It has the potential to rapidly alter well-paid white-collar jobs, and that's something we have little choice over. However, we can experiment with the technology and learn new ways to do things. The speed at which these changes are happening means that they will affect strategic decisions, and boards will need to respond accordingly. One of the major issues is the disconnection between those who understand AI and those who make strategic decisions. This has the potential to lead to some very unfortunate consequences, not just for the business but for society. Any rapid adoption of AI that leads to higher unemployment will be socially disruptive, but that's not all. Many employees find fulfilment in their work, and if you automate the creative and interesting aspects of what they do, you risk adversely affecting morale, reducing engagement, and creating a weaker culture. But this doesn't mean ignoring what's going on. AI can increase productivity and creativity, and if it can be successfully applied to your business, market competitors will threaten your long-term viability if you don't utilise it. I know I struggle with AI because it has seemingly come from nowhere, and it has just been let loose on the public, largely as a result of having a natural language interface (just like ChatGPT). I don't think I'm alone in discovering that it's a whole lot better at doing things than expected, and as every day passes, I am more impressed. If we assume that we are all starting from a low base, how do you build up the mental muscle that allows you to understand enough to make sensible strategic decisions? Part of the answer is surely experimentation. If you try ChatGPT you will find it's straightforward, and you can learn some practical skills just by varying the prompt. Taking this a step further, you could build a safe space where you could collectively work with colleagues to improve executive AI fluency. This is not the same as coding; it's about developing a mental mindset that gives you insight into what AI can do for your business. Practical suggestions could include taking a document and repurposing it, asking it to explain some data trends, and building a new advertising campaign but don’t use proprietary data until you are clear about where it goes. It doesn't take a lot of imagination to see that just about every part of a board's information deck could, in the future, be AI-generated and nuanced for the reader, who would have the ability to summarise and drill down as required. But that's only the start. In the industry I'm familiar with, financial services, there's a huge amount of proprietary data held across a multitude of systems. That data is frequently in documents, contracts, and spreadsheets, making it very hard to access. There's also a deep history of legacy data that resides somewhere, and all of this is held together with tape and string. Now, with AI this data becomes extremely powerful. It can help you price and manage risk like it's never been done before. Underwriting credit and insurance risk immediately springs to mind. There's also a separate business case that takes data and sells access to it through an interface like a chatbot. This may sound fanciful, but it's something that's already being done by firms like Bloomberg. These two business models won't suit everyone, but there is a third way: to pick the low-hanging fruit and implement AI as and when it becomes commercially available from third parties and embedded in systems and applications that you already use. Examples include using chatbots for customers and implementing fraud detection systems to identify irregular patterns of behaviour. How you go about this depends on whether you can make smaller incremental changes leading to increased efficiency or if you are pushed into much larger, rapid changes to respond to competitors. There are also limitations to what can be achieved. In a regulated industry, you need to be aware of what data you can use and how you can use it. If you get this wrong, there are serious financial and reputational risks; therefore, banking cannot be as entrepreneurial as many lightly regulated businesses. Furthermore, regardless of the path you take, the results are only as good as the information used in the model, and we know that a lot of banking data has been subject to manual intervention which embeds errors. It also seems evident that if decision-makers can't explain how a model is trained, how it works, and what its limitations are, the regulator is not going to be happy. This is particularly important when considering GDPR and the principles of treating customers fairly, as well as the application of complex risk management techniques. There's no doubt that enterprising employees are already using AI in their personal workspaces, and for better or worse, it will leak into your business. Providing them with guidance on what is acceptable is a good place to start. Overall, AI experimentation and understanding are essential in navigating the changing landscape of knowledge work. By embracing the potential of AI, businesses can increase productivity, creativity, and strategic decision-making. However, it is crucial to consider the potential consequences, such as job displacement and employee morale, and find ways to mitigate them. With proper experimentation, collaboration, and a thoughtful approach, AI can become a powerful tool to drive growth and success in the business world.

9 views0 comments

Recent Posts

See All

Comments


bottom of page