Navigating the Digital Transformation Maze: Challenges, Costs, and AI Opportunities

I’ve been thinking a lot lately about the impact of digital transformation on governments and businesses and wanted to continue my recent spurt of posts to help my network with a taste from Gartner's experiences in the last year. It's fascinating to see how organisations are using technology to improve efficiency, citizen experience, and even mission outcomes and it's worth jotting down some notes on the utilisation of AI.....what are others finding when they have a go.

There is a great deal at stake, and the stakes continue to rise. Government digital transformation initiatives are notorious for their high failure rates, often resulting in dramatic setbacks, as seen with projects like NPfIT or Horizon. One of the most significant threats to these initiatives is cost, which can be especially burdensome in today's economic climate. Our analysis and research have revealed that more than half of organisations abandon their efforts due to grossly inaccurate cost estimations. In some cases, estimates can be off by 500% to 1000%, representing a staggering margin of error.

Generative AI presents a genuine opportunity for governments and businesses to gain a operational improvement. However, there's still considerable hype surrounding its actual capabilities. Let's delve deeper into the reasons behind this and the numerous challenges that need to be addressed. As I stated earlier, one of the primary challenges remains the cost. Implementing and maintaining Generative AI can be exceptionally expensive. It's crucial for organisations to carefully evaluate the costs and benefits before making any decisions. Though it might seem obvious, it's worth highlighting that there are actually eight key cost factors that are particularly volatile and can determine the success or failure of a project. These include AI data, model, development, and token costs, among others. Environmental impacts and costs are another thorny issue.

Another challenge is data readiness. GenAI models (or any AI/ML model) need to be trained on large amounts of quality data not rubbish stuffed down the back of a few excel sheets but structured, curated content. Organisations need to ensure that they have the right data in place before they can start using AI in any shape.? And if you're using language models you better have wired brushed them thoroughly before setting them free into the wild.

Gartner are seeing an emerging market focusing on integrating fragmented data management components with data fabric and generative AI capabilities. This is a good thing as it promises to reduce your costs and technical debt. GenAI and active metadata management can help you automate various manual tasks in data integration, though strong metadata management is crucial to maintain accuracy and prevent issues like hallucinations. But then this leads to another challenge - your people. Most don't have the skills in place to curate and modify their existing data so best you start upping your training game, skills inventory and AI/data literacy. Worse still departments argue as to who 'owns' the data, technology and processes. I'd say the business does on behalf of your citizens.

Even if you get the cost and data piece right, there can be productivity challenges. A recent study of GenAI for investment services found that an Advisor Co-Pilot actually performed worse than an Human Advisor equivalent.

So, what can organisations do to make sure that their digital transformation initiatives are successful? Well assuming you've already identified and agreed on the relevant use cases:

• Firstly, measure and report on the contribution of digital solutions. Digitally advanced governments are much more likely to measure impact in areas such as mission outcomes and regulatory compliance.
• Secondly, embrace contemporary practices such as Agile project delivery, human-centred design, and DevSecOps. For instance, only 15% of less-advanced government organisations have incorporated human-centred design, compared to half of digitally advanced governments.
• Thirdly, build a composable technology architecture that can support the integration of different technologies and platforms. Digitally advanced governments globally are deploying a wide range of technologies, including AI, chatbots, and machine learning.

Digital transformation is a journey, not a destination. Organisations need to be constantly learning and adapting to new challenges and opportunities. By taking a strategic approach to digital transformation and addressing the challenges head-on, organisations can reap the many benefits of digital transformation and stay ahead of the curve. Just don't under estimate the size of the task coming your way.