IPA Director on new AI project delivery framework

The Infrastructure Projects Authority (IPA) is the UK’s centre of expertise for infrastructure and major projects. Their newly released framework, Data Analytics and AI in Government Project Delivery, sets out how AI will be experimented with and harnessed by government project delivery professionals to optimise the opportunities of AI and data for better project outcomes.

Listen to 'IPA Director on new AI project delivery framework'
6:25


To find out more about how this framework will work in action, we spoke with Karina Singh, Director of Function at the IPA. Read until the end to hear closing remarks from Baroness Neville-Rolfe, Cabinet Office Minister of State.

Q: How can the New Framework be applied across departments with varying AI maturity?

The framework will support all departments, irrespective of their maturity. Rather than being something that needs to be ‘applied’, the framework sets out actions that we’ll take centrally to empower and enable departments in using data and AI. 

As a result, departments have the flexibility to focus efforts on what’s important for them as an organisation or a team. The framework empowers them to go forward with the work at whatever level they’re currently at, with the confidence that they will be supported in finding solutions to the challenges they face in their projects and programmes.



For example, by defining a common set of data standards and improving the interoperability of platforms and tools, departments will be able to derive better insights from their data and the data of others. 

Similarly, by steering experimentation towards the most broadly applicable and value-adding use cases, we can help all departments boost performance across the board.

Q: Your framework has highlighted that your initial approach will be 'no regrets' low risk innovation. What criteria will be used to define the low risk threshold.

By low risk, we mean focusing our efforts on the most broadly applicable scenarios, being prepared to ‘fail fast’ or adapt to changing knowledge/circumstances, and being very careful about investment in new, high-cost infrastructure until we’re more certain of the benefits. Initially, given the risks of not understanding the tools and flaws in the data we are using, the focus will be on ‘no regrets’, low risk innovation, in limited environments with checks and balances, to drive efficiency and productivity. 

As our understanding and confidence improves, we will become more advanced in using data analytics and AI to improve delivery, continuing to concentrate on applications that contribute to wider government priorities and benefit the UK and its citizens.

Q: One of the features of your approach will be experimenting and innovating at local levels to generate actionable insight and ways of working. Can you elaborate on any future plans for experimenting and innovating with local government?

We recommend that the starting point for experimentation and innovation should be with the project delivery community  across central government, tackling the challenges specific to their projects and programmes. A bottom-up approach will support project teams in learning lessons from experiments on low-risk opportunities, before expanding solutions to other areas.

To enable this, the IPA is participating in ‘hackathons’ run by organisations such as Evidence House, led by the 10 Downing Street Data Science team. These events bring together data scientists, engineers and coders from across government and academia to spend focused time experimenting, generating evidence-based data analytics and AI solutions to priority problems across government.

Given the common issues faced and the interconnected nature of the profession, these approaches apply equally to both central and local government.

Q: One of the core objective of the Government Project Delivery Function Strategy is 'influential leadership'. The Association for Project Management has highlighted that once of the potential impacts of AI in project management may be reducing the sense of achievement. How can departments foster this whilst still making use of AI to optimise project outcomes? 

The most prominent influence of data analytics and AI on existing project delivery roles is the need for staff to upskill. IPA is working to ensure that data and AI skills are incorporated into our accreditation scheme and project delivery learning offer to ensure colleagues are equipped for the opportunities posed by AI.

Some roles may change and be re-defined to include data analytics responsibilities and competencies, while entirely new roles are also likely to emerge. By capitalising on our world-leading comprehensive learning and development offer and harnessing data analytics and AI skills, civil servants working in project delivery can gain a new sense of achievement, adapting to a changing environment.

We will continue to share successes across the community, showcase the productivity improvements that data analytics and AI bring to projects and reward more predictable outcomes.

Q: How does the IPA hope to organise their timeline to implement priorities across government?

We will iterate and be agile in our approach as our understanding of the opportunities and risks grow. Progress across all areas will not be linear, but will be informed by access to information, resources and tools.

Baroness Neville-Rolfe, Cabinet Office Minister of State:

The report shows how collaboration is at the heart of exploring possibilities of using AI in government projects - those that can deliver critical, transformative and life-changing public services. By working across government and with wider stakeholders, the IPA is laying a path for the future of project management in the UK - one that is not only making innovative use of AI in government, but also sets best practice for wider industry.  By promoting a more open and collaborative culture and experimenting with new ideas in a responsible way, we can build a thriving, innovative and bright future for the United Kingdom.

New call-to-action

Also Read