Output based requirements should be a key factor in procuring artificial intelligence solutions for public services, according to new government backed guidelines.
The Office for Artificial Intelligence (OAI), a joint unit under the Department for Business, Energy and Industrial Strategy and the Department of Digital, Culture, Media and Sport, has published the guidance as a summary of best practice along with insights on dealing with some specific challenges.
“As AI is an emerging technology, it can be more difficult to establish the best route to market for your requirements, to engage effectively with innovative suppliers or to develop the right AI-specific criteria and terms and conditions that allow effective and ethical deployment of AI technologies,” it says.
The document points out that AI technologies are developing rapidly, which makes it more difficult for buyers to get a clear view of the market anticipate which will be the right solutions for their purpose.
This makes it important to include output based requirements in the tender document, describing the challenges faced and any opportunities that have been identified. This will allow suppliers show which technologies could be most appropriate.
Statement and strategy
Key considerations in drafting the requirements are to start with a problem statement, highlight data strategy requirements, focus on data quality issues and underline the need to understand the supplier’s AI approach.
There is also a need for careful preparation and planning before the procurement, involving multidisciplinary teams to establish whether the use of AI is viable, and to lay the ground for data assessment and governance.
The teams can also contribute to the selection process, which needs a broad base of experience and expertise in what the organisation needs to achieve. Along with this is a need to look at the skills, qualifications and diversity needed in the team that will deploy any system.
For contract implementation and ongoing management, the OAI advises the use of process based governance and auditability, model testing, knowledge transfer and training and a consideration of what the end-of-life processes and data should look like.
The document sums up some core principles for AI procurement:
- Include the procurement within a strategy for AI adoption.
- Make decisions in a diverse and multidisciplinary team.
- Conduct a data assessment before beginning procurement.
- Assess the benefits and risks of AI deployment.
- Engage with the market from the outset.
- Establish the right route to market and focus on the challenge rather than a specific solution.
- Develop a plan for governance and information assurance.
- Avoid Black Box algorithms – those which are almost impossible to understand – and vendor lock-in.
- Focus on technical and ethical limitations during an evaluation.
The guidance was developed with the World Economic Forum for the Fourth Industrial Revolution, the Government Digital Service, Government Commercial Function and Crown Commercial Service.
The OAI says in the document that it is not exhaustive and more will follow. Last year it published a guide on the use of AI in the public sector, emphasising need to pay careful attention to data quality, fairness, accountability and privacy.
Image from iStock, Andrey Popov
