Include outcome-based requirements when tendering for AI

posted by Marino Donati
15 June 2020

Public sector buyers looking to procure artificial intelligence (AI) should ensure they include clear output-based requirements when drafting an invitation to tender, according to UK government guidance.

In a guide the Office for Artificial Intelligence at the Department for Digital, Culture, Media & Sport said procurement should start with a sufficiently detailed problem statement clearly setting out challenges and opportunities being faced, to allow suppliers to properly determine the most appropriate technologies.

The document said because AI is an emerging technology, it can be more difficult to establish the best route to market, to engage effectively with innovative suppliers or to develop the right terms and conditions that allow effective and ethical use of AI technologies.

The guidance is aimed at central government departments who are considering the suitability of AI technology to improve existing services or as part of future service transformation. Other public sector bodies may also follow the guidelines.

As well as a problem statement, tender publications should describe how the AI system will fit within current data strategy and practices, detailing current limitations and requesting potential suppliers to detail how they would address shortcomings.

Buyers should also underline the need for an “explainable approach” to AI development so that outputs from the AI system will be able to be interpreted by the procurement team, making it easier to engage with other suppliers in the future and avoid vendor lock-in.

There should also be mutually beneficial and fair arrangements about ownership of any new intellectual property generated by new AI systems, the guidance said.

During the selection process, the experience of a multidisciplinary team should support the evaluation process and ensure there is a broad base of expertise.

 Want to stay up to date with the news? Sign up to our daily bulletin.

CIPS Knowledge
Find out more with CIPS Knowledge:
  • best practice insights
  • guidance
  • tools and templates