Public sector procurement — Practical impact of AI

Susan Hilgers from Proxima and Ian Makgill of Spend Network discuss some of the practical impacts on public sector buyers and suppliers that the increasingly used new technology is having

In the rapidly evolving landscape of business technology, one particular evolution is leading the conversation: generative artificial intelligence (AI). This cutting-edge branch of AI harnesses vast datasets to create dynamic content for its users, from text to images and videos. In a 2023 survey from Forbes Advisor, 64% of businesses expect AI to increase their overall productivity, demonstrating the increasing confidence that generative AI will transform business operations. 

Susan Hilgers from Proxima and Ian Makgill of Spend Network discussed some of the practical impacts on Public Sector Buyers and suppliers that the increasingly used technology is having.

We are hearing that one of the great benefits for suppliers is the ability to quickly generate good responses to standard tender questions. Great for small and medium-sized enterprises, including voluntary, community or social enterprise organisations, who do not have huge departments of bid writers and who cannot afford external help responding. So potentially a huge plus for accessibility to public spend.

This increase in use does create a few issues for commercial teams and evaluators.  More tenders that appear credible and compliant, may make it hard to differentiate and may create more bid bunching. The Buyer needs to validate the responses and check that the descriptions provided are true and examples real as the process must be fair to all bidders. The buyers may also be faced with many more responses than before but no increase in their commercial and evaluation resource. How will they maintain quality and timeliness?

Some private sector buyers are using technology to reverse check and potentially exclude those written using AI.  For the public sector this would create a real risk as the checkers can fail to identify and falsely identify, meaning you could potentially both fail to qualify and fail to disqualify. Also is it proportionate to exclude a supplier because the response has leveraged a digital tool? Is it wrong for organisations to drive quality and efficiency improvements in bid writing by using AI? Falsify information, yes. But using tech to optimise their own business processes? Bear pits for the buyer here.

Practical steps for commercial teams to consider:

  • Badly structured questions can play into the hands of auto bidders, making it hard for good tenders to excel. Are you drafting your criteria and questions well i.e. in a way designed to elicit great responses to your requirements? If you are posing a smaller number of great questions rather than an “everything but the kitchen sink” approach, then the bidder is less likely to generate boilerplate answers and more likely to use their knowledge and know how to suggest great solutions and give you real examples and proof of ability.
  • Consider carefully which questions could be answered yes/no and checked as part of assurance and which do not need to be answered elaborately and scored, freeing both you and your bidders to focus on what differentiates and what adds value and validating that.
  • Revisit your bid validation process – is it good enough to identify bids that are well written but baseless? Do you over rely on the written word? Should you be including more “show me” into your procurement for example solution walk throughs, demos, references as well as the “tell me” elements of the written response and paper/digital evidence.
  • Revisit how you set out the competition rules – if you validate bid elements later in the process, and potentially only the top scoring bidders, then should you include a statement of what will happen in the event one or more of the top scorers fail validation?
  • Look at your timescales & resources– will you need to build in more to evaluate and check more bids?

Could the buyer leverage AI to automate evaluation and speed up that process?  Our advice is do not attempt without legal advice as this may open you to procurement risk. Public Sector buyers need to be able to set out the reasons for award and rationale for each score, an AI evaluation is unlikely to be risk free with the current state of user training and awareness.

It’s a fast-moving situation and there is untapped potential for commercial teams elsewhere, for example digesting complex contract performance data into a diagnostic report to support contract management. We will cover other potential uses in our next article.


Read the most recent articles written by Proxima - Preparing for Procurement Reform: Three proactive steps


Share this page