Mastering The Challenges Of Data In Local Government | Peak Indicators
Understanding how to source and build high-quality data sets is crucial for creating insights that councils can trust. Data must be accurate, timely, and offer the most complete view of the world for it to reliably support decision-making.
Paul Clough of Peak Indicators explains the tools and techniques councils can use to build the most reliable insights and what they can learn from previous, high-profile failures in data-quality.
Catch the full podcast episode
Paul Clough, Head of Data Science and AI
Not just in local government, but in industry as a whole, there is this assumption that everybody is way ahead — that everybody is ready to be using data science, machine learning, predictive analytics, and so on. But I think the reality is that many organisations, not just local governments, are actually struggling just to get the data that they have under control.
We have the challenges of data being siloed, data being distributed across many legacy systems, across many kinds of use cases — different bits of the organisation. And it’s not just a technical problem. Sometimes it can be a cultural problem as well, when people hang on, or are very precious about certain data that they have and data sources and they don’t necessarily want to share that. Even if it could improve analytics or services.
I think understanding how data flows — so what we might call data lineage — and being able to track data flows is really important as well. Particularly now for GDPR, where we need to know where the data is going to, what systems the data is being used by, what reports are using particular data sources.
Nathan Makalena
How can councils mitigate AI bias? And what sort of lessons can be learnt from some of the past, high-profile failures using the technology?
Paul Clough
Tackling some of these issues is super important. I think along with bias, we need to be really thinking about ethics. The ethics of data that we have and the usage, but also the ethics of AI. The starting point has to be asking yourself, what are we doing with this technology? Why are we using it? Where could these biases creep in? What might they look like? And then you can start to tackle that.
Often the biases are in there because of the biases in those human processes and decision making. Humans generate the data, humans create the algorithms and humans make use of the output, therefore there will be biases. What we need to be doing is just be realistic about that and accept it, understand it, then think about, ‘okay, what’s the implication?’ We need to make our processes transparent, so that we have trust: trust in the approaches that we use for data collection, trust in the approaches we use for analysing that data, and presenting that data and ultimately using the data.
Now, those human processes should also be transparent in terms of how a decision is made, particularly when it affects individuals — when it affects citizens. Local government is in a great position because they’ve been doing this and thinking about these things for a long, long time. I guess they’ve just got to be thinking about this with respect to new technologies, and just how they transform themselves and transform their processes and their organisation to engage with these and make use of them in very effective ways.
Originally published at https://www.peakindicators.com on December 16, 2022.
