Procurement and Artificial Intelligence – how to approach the use of AI in the procurement process
Despite what the Government had previously set out in its Green Paper on transforming public procurement around the key driver of innovation and the use of innovative procurement practices, the Procurement Bill itself is completely silent on the use of Artificial Intelligence (AI) as part of a procurement process.
Whilst AI continues to evolve, it is becoming increasingly wide spread, so what do practitioners need to think about now in order to envisage this fast-moving agenda in their procurement activity
Addressing AI in tender documents
It is widely accepted that a procurement practitioner is unlikely to be able to stop a bid team from using AI in the preparation of its response. Even if the tender documents set out that the use of AI is not permitted, this will be difficult to monitor in the course of a procurement.
Instead, it may be worth requiring bidders to declare where they have used AI as part of their bid response (for example, which question they have used an AI platform for). Whilst this doesn’t stop the use of AI in a tender submission, it will at least allow clients to identify and understand where AI is increasingly relied upon, and may help to shape questions differently going forward.
Contracting authority's should also understand the risks around confidential information being fed into an AI platform, with the consequence that data / information is disclosed beyond its intended recipients. This will largely depend on the type of platform that a bidder might use (for example, whether it is a closed offline platform, or an open online platform), but the tender documents will need to set out clearly the approach to be adopted in respect of confidential information.
Bidders will need to understand what is / is not considered confidential, and it is therefore recommended that where a contracting authority considers confidential information should not be fed into an AI platform that this is explicitly set out in the tender documents. The contracting authority will then need to conduct an exercise of identifying what is considered "confidential" and this will need to be communicated clearly to all bidders.
Another push factor which might require contracting authorities to take a proactive approach to the use of AI is the way in which quality submissions tend to be evaluated under public procurement tender exercises.
With an increased use of AI, now might be a good time to consider whether the common approach to quality evaluation is no longer the most appropriate (i.e. are exam questions now the best way to evaluate quality proposals, where responses to those questions may be generated via an AI platform?).
Instead, contracting authorities should be thinking now about how they can craft project specific questions which require a delivery team's input, and which will generate more considered responses which align to a contracting authority's own requirements and objectives.
It is clear that nobody has all the answers to these issues at present, and as the use of AI (and the AI platforms themselves) continue to evolve, contracting authorities and bidders alike will need to be mindful of various key issues in their approach to procurement.
We will consider some of those key issues in the coming weeks in a series of AI focussed procurement insights, including:
- How can a contracting authority procure AI?
- How to address data and AI
- The interaction between AI and public law considerations
- The benefits of AI