Since the launch of Chat GPT, companies have been keen to explore the potential of generative artificial intelligence (Gen-AI). However, making the most of the emerging technology isn’t necessarily a straightforward proposition. According to Carruthers and Jackson Data Maturity Index, as many as 87% of data leaders said AI is either only being used by a small minority of employees at their organisation or not at all.
Ensuring operations can meet the challenges of a new, AI focussed business landscape is difficult. Nevertheless, organisations can effectively deploy and integrate AI by following steps. Doing so will ensure they craft effective, regulatory compliant policies, which are based on clear purpose, the correct tools and can be understood by the whole workforce.
Rubbish In Rubbish Out
Firstly, it’s vital for organisations to acknowledge that Data fuels AI. So, without large amounts of good quality data, no AI tool can succeed. As the old adage goes “rubbish in, rubbish out”, and never is this clearer than in the world of AI tools.
Before you even start to experiment with AI, you must ensure you have a concrete data strategy in place. Once you’ve got your data foundations right, you can worry less about compliance and more about the exciting innovations that data can unlock.
Identifying Purpose
External pressure has led to AI seeming overwhelming for many organisations. It’s a brand new technology offering many capabilities, and the urge to rush the purchasing and deploying of new solutions can be difficult to manage.
Before rolling out new AI tools, organisations need to understand the purpose of the project or solution. This means exploring what you want to get out of your data and identifying what problem you’re trying to solve. It’s important that before rolling out
AI, organisations take a step back, look at where they are currently, and define where they want to go.
Defining purpose is the ‘X’ at the beginning of the pirates map, the chance to start your journey in the right direction. Vitally, this also means determining what metrics demonstrate that the new technology is working.
The ‘Gen AI’ Hammer
While GenAI has dominated headlines and been the focus of most applications so far, different tools and processes are available to businesses. A successful AI strategy isn’t as simple as keeping up with the latest IT trends. A common trap organisations need to avoid falling into is suddenly thinking Gen AI is the answer to every problem they have. For example, I’ve seen some businesses starting to think… ‘everybody’s got a gen-AI hammer so every problem looks like that is the solution you have to use’.
In reality, organisations require a variety of tools to meet their goals, so should explore different technologies, but also various types of AI. One example is Causal AI, which can identify and understand cause and effect relationships across data. This aspect of AI has clear, practical applications, allowing data leaders to get to the route of a problem and really start to understand the correlation V causation issue.
It’s easier to explain Causal AI models due to the way in which they work. On the other hand, it can be harder to explain the workings of Gen AI, which consumes a lot of data to learn the patterns and predict the next output. There are some areas where I see GenAI being highly beneficial, but others where I’d avoid using it altogether. A simple example is any situation where I need to clearly justify my decision-making process. For instance, if you need to report to a regulator, I wouldn’t recommend using GenAI, because you need to be able to demonstrate every step of how decisions were made.
Empowering People Is The Key to Driving AI Success
We talk about how data drives digital but not enough about how people drive data. I’d like to change that, as what really makes or breaks an organisation’s data and AI strategy is the people using it every day.
Data literacy is the ability to create, read, write and argue with data and, in an ideal world, all employees would have at least a foundational ability to do all four of these things. This requires organisations to have the right facilities to train employees to become data literate, not only introducing staff to new terms and concepts, but also reinforcing why data knowledge is critical to helping them improve their own department’s operations.
A combination of complex data policies and low levels of data literacy is a significant risk when it comes to enabling AI in an organisation. Employees need clarity on what they can and can’t do, and what interactions are officially supported when it comes to AI tools. Keeping policies clean and simple, as well as ensuring regular training allows employees to understand what data and AI can do for them and their departments.
Navigating the Evolving Landscape of AI Regulations
Finally, organisations must constantly be aware of new AI regulations. Despite international cooperation agreements, it’s becoming unlikely that we’ll see a single, global AI regulatory framework. More and more, however, various jurisdictions are adopting their own prescriptive legislative measures. For example, in August the EU AI Act came into force.
The UK has taken a ‘pro- innovation’ approach, and while recognising that legislative action will ultimately be necessary, is currently focussing on principles-based, non-statutory, and cross-sector framework. Consequently, data
leaders are in a difficult position while they await concrete legislation and guidance, essentially having to balance innovation with potential new rules. However, it’s encouraging to see data leaders thinking about how to incorporate new legislation and ethical challenges into their data strategies as they arise.
Overcoming the Challenges of AI
Organisations face an added layer of complexity due to the rise of AI. Navigating a new technology is hard at the best of times, but doing so as both the technology and its regulation develops at the pace that AI is currently developing presents its own set of unique challenges. However, by figuring out your purpose, determining what tools and types of AI work and pairing solid data literacy across an organisation with clean, simple, and up to date policies, AI can be harnessed as a powerful tool that delivers results, such as increased efficiency and ROI.
- Data & AI
- People & Culture
