The hard work of change, hype cycles and why LLMs aren’t a quick fix

There has been a tsunami of hype recently about Large Language Models (LLMs) such as ChatGPT, Bard and so on. To me it has felt quite similar to previous hype cycles, such as with blockchain – “the end of banks” vs “the end for programmers”. For a long time I would get frustrated with people leaping into the latest ‘hot’ technology because I felt they weren’t understanding the hype cycle nor the complexities of how technology really works. However, now I think something more fundamental is going on: In essence, people are consciously or subconsciously, trying to find ways around the long slow hard work of delivering fundamental change (and for me this is specific to work in public services).

Just chucking in some extra technology doesn’t deliver genuine change. Through decades of hard-won experience we know that technology-led change just does not work. It’s only through multidisciplinary teams working in a user-centred way iterating on user feedback that genuine, lasting improvement happens — it is culture change working in step with technology. This is the way set out in the UK service standard, through which we have been able to fundamentally reimagine (some) services and make a positive difference.

I very much know there’s still such a long way to go on the change journey. It is hard yards and we are at the very beginning. It can’t be led by technology, it’s about people and making a difference. When there is so much legacy tech, with poor data models around, I do really understand the wishful thinking that something new could skip the pain of sorting it all out. I know, I feel the pain. But actually doing the hard yards of building the right culture, the right data structures and the right services is what needs to come first.

The tools, technologies and the connectivity of the Internet (a la Loosemore) have allowed us to do public services in fundamentally different ways with a very different cost model, but that alone is not enough. So adding the newest hype technology will never leapfrog lasting change of our culture, behaviours, and imaginations. Indeed, I think the most important shift technological change has delivered is how it has opened our minds to genuinely re-imagining public services for the better. And that is the work. Let’s go.

For further reading on how LLMs work, and how to think about them, I recommend ChatGPT is a Blurry JPEG of the Web and the lengthy What is ChatGPT Doing … and Why Does It Work? followed by this UK government guidance.