Another day and yet another government technology scandal; this time, it’s our friends at Deloitte who have been caught out with their hands in the AI cookie jar.
Deloitte has admitted to using generative AI to produce a $440,000 report that contained several errors including incorrect references and citations. Deloitte was commissioned to produce the report for the Department of Workplace Relations, looking at the IT system used to automate penalties in the welfare system.
Why colour me surprised! Imagine the cheek of it – a technology service delivery firm actually using technology to make their work easier.
How dare they! And on the public dime!
As bemusing as I find this state of affairs, I need to make two critical points:
- Australians need to make a significant adjustment to the way we think of AI. Just because something is AI does not make it automatically bad. Being afraid of AI in this late stage of 2025 is like being afraid of the internet in 2005. Get with the program or find yourself another planet.
- Technology service provision in government should never have been outsourced. It was a bad idea in the 90s and still remains a bad idea 30 years on.
Technology in government is not optional; technology is the skeleton, muscle, sinew and skin for the delivery of all services across federal, state and local government lines. You cannot run a simple farmer’s market or a council pet registration service without technology. Leaving government enabling technology at the behest of consulting firms exposes us all to an unacceptable level of sovereign risk.
Don’t believe me? Let’s break it down with a single example. Let’s take a government service dear to all Australians’ hearts: Medicare.
Medicare has more personal information and transaction data housed in data warehouses than most of us can imagine. It has relational data (who is someone’s parent and what their family structure is), personal identifying data (who are you, medical data of what tests you have) and transactional data (how much you have paid to medical services).
Now imagine all of that information not being properly recorded and maintained and, worst of all – breached. I am not engaging in hyperbole here – we’ve seen plenty of data breaches in the last 3 years to know that this is in realms of possibility. But what if I were to tell you that data breaches and technology failures happen more as a function of neglect than wilful malfeasance?
This is because building technology is not like building a house or a bridge. It’s been treated like any other engineering project – this is why governments have felt safe to get external contractors to build technology – but technology development is materially different to construction development. Technology development is ever evolving and having a revolving door of technology service providers means that the intellectual property walks out the door the minute the project has been delivered and handed over to the client.
Think of it this way: you’ve built a house and you hand it over to the owner, except the owner has no idea how to maintain said house or worse, the people who commissioned the house have left and you are now the third owner down from the initial build. And the materials that were used in the build have now been discovered to have asbestos. Besides, the people best placed to help you with the maintenance of the house have moved onto the next project and no one knows what language was used to code the light switch or where the plumbing is.
Surely the documentation – the blue prints – of the system ought to tell you how to maintain and support this house? What if I were to tell you that no one ever reads those documents that are often between 500-7,000 pages long. No one.
But the test scripts and the users? Yeah – the testers left with the technology service providers as soon as the hypercare period expired, and since everyone is mad that no one delivered what was promised, no one is talking to each other.
I have watched both public and private sector IT projects fall apart from multiple angles; from eCommerce projects as a young graduate to large scale technology transformations as a product lead – and the one thing that management continues to fail to grasp is that you cannot stop investing in technology. The minute a technology project is delivered, it is rendered obsolete.
This is why I have been a proponent of the Scaled Agile Methodology which has long standing teams at the heart of its delivery framework. The people who build the system are those who maintain it and know where the bodies are buried. The development team is not stood down at the end of the delivery of the ‘project’ but become guardians who ensure that it is well maintained and ensures it is then upgraded when the time is right into a newer technology stack.
The Scaled Agile Methodology also limits risk and ensures that there is an owner for the technology product. That means you know with whom the buck stops with and no one is running around trying to find the person with the most knowledge of the system
Governments and businesses will no longer be required to undertake massive risk-generating technology upgrades when the maintenance and continuous upgrades have meant the expiry date is well managed and transition is faultless.
The Australian government must stop dolling out billions on technology projects to consulting firms. More often than not, they have graduates with little to no practical knowledge running these projects and they haven’t the experience to deliver a letter, much less core government-enabling technology. We need to build the capability inhouse and that means having a large development capability.
And you know what comes from having large groups of people working on technology projects?
Innovation.
Now imagine that – having Australian government owned innovation that could drive our economy. Instead of sending offshore the profits of consulting firms; we develop our own capability and we own the profits of it.
That is the technology innovation pathway I can get behind.