The following is an example of what can be generated by AI with minimal input from the bidder.

While the above provides a reasonable generic response to the Fair Work and Equality aspects of the request, there is no substantive detail as to how the catering company in the example would deliver the service.

This illustrates the fact that bidders should not rely solely on AI to produce a bid. It is always essential that there is human intervention and contribution by bidders prior to submitting their bids.

You should inform bidders that it is their responsibility to check their bid before it is submitted to ensure accuracy and completeness.  Proportional due diligence should also be conducted prior to contract award. 

As always, you should aim to write your tender documents in a way that will result in clear, tangible, responses from your bidders.  This will mitigate against bidders trying to solely use AI generated responses, which tend to be generic and are unlikely to be sufficiently tailored to specific requirements outlined in the ITT.

Artificial Intelligence (AI) may have an impact on how buyers and bidders function. Below are some of the key considerations for AI in Public Procurement.

Sensitive Information

AI systems share and use information and data that users input e.g. into AI tools or any systems that make use of AI for any purpose.  As a result you should be aware of this and actively manage this risk of sensitive information or copyrighted material being shared inappropriately.

When preparing tender documentation, you should always ensure that you do not add:

• personal data
• unpublished/restricted data or
• data which might suggest future government or public body activity (which may not be in the public domain)

into any AI tools or into any systems which make use of AI for any purpose.

You should also seek advice from data security colleagues in accordance with any internal governance regarding any potential risks of data security breaches associated with any AI data platform you intend to use. 

You should ensure that bidders are made aware that any information that is personal, classified, sensitive or reveals the intent of government or the public body (which may not otherwise be publicly available) included in any tender documentation must not be input by bidders into any AI tools.

If AI can Potentially be Used in the Delivery of  the Service

Buyers should engage with stakeholders in the development of the specification and tender documentation if AI could be used in the delivery of the service.  This should be done in accordance with your organisation's policies and guidance and any associated standards or quality checks that may be required.  Input from colleagues such as AI data policy experts, cyber security or legal advisers is advised in this decision-making process, determining required quality standards, which may require human checking and considering reliance on any particular supplier.

Bidders Using AI to Generate Bids

As the availability and functionality of AI tools grow, bidders may consider, or may already be using AI (including generative AI) to generate their bids.

AI can be used by bidders to help with the initial drafting, formatting and the general quality of bids.

AI may also enable bidders who have the relevant capability and capacity to bid (especially SMEs and third sector businesses) when they would have previously prioritised their bid team resource elsewhere.

However, you should be aware that AI may sometimes also generate inaccurate or misleading statements.  Statements, facts or references generated can appear plausible, but are in fact false. This is because generative AI will gather all information, whether fact or fiction, current or outdated, so it can be inconsistent, misleading or inaccurate.  Generative AI will also try to return the answer or output it thinks you are looking for.  So it is not deliberately misleading, but it can propose information which it has pulled from looking at other similar examples or bids.

Bidders who do not check their bid carefully could accidentally include information which is not accurate because of poor checking rather than as a method or attempting to deliberately mislead.

Whether the response is prepared by the bidder themselves, a bid writer, a consultant or with the aid of generative AI, bidders should be made aware in the tender documentation that it is their responsibility to check their bid before it is submitted to ensure accuracy.

You should consider that, if you expect AI to be used in generating bids, you may wish to carefully think about how you will evaluate very similar bids (although bidders should, of course, be tailoring their bid for the specific procurement).

You may wish to consider and plan, where appropriate, for a general increase in procurement activity.  This is because bidders may use AI to streamline or automate their processes and improve their bid writing capability and capacity, which may lead to an increase in the number of bids you receive.

Equal Treatment, Non-Discrimination and Evaluation

You must adhere to the principles of equal treatment and non-discrimination, irrespective of whether bidders have used AI to generate their bid or not.

As with all bids, buyers should ensure proportionate due diligence is conducted prior to contract award and throughout the duration of the contract, to ensure bidders can deliver the goods and services e.g. clarifications may be required as appropriate.

Considerations for Buyers Using AI

The Central Digital and Data Office (CDDO) leads the UK Government Digital and Data function.  They have produced guidance which should be used by civil servants in relation to their use of AI.

The Scottish Government also advises that personal or sensitive information should not be used in AI tools and promotes this UK Government guidance for civil servants on the use of generative AI.

If you are not a civil servant, you should check if your own organisation has AI guidance/governance procedures which you should use.

As AI is a fast changing area you should ensure you keep up to date with the most relevant guidance.  Like bidders and other users of AI, you will need to check the accuracy of any AI generated information that you use.  You should be aware that AI may sometimes generate inaccurate or misleading statements; where statements, facts or references appear plausible, but are in fact false.  Generative AI will gather all information, whether fact or fiction, current or outdated, so can be inconsistent, misleading or inaccurate.

An Example of AI Being Used in a Bid

The following is an example of what can be generated by AI with minimal input from the bidder.  This same example is also shown on An Example of AI Being Used to Write Bid Content.

While the above provides a reasonable generic response to the Fair Work and Equality aspects of the request, there is no substantive detail as to how the catering company in the example would deliver the service.

This illustrates the fact that bidders should not rely solely on AI to produce a bid. It is always essential that there is human intervention and contribution by bidders prior to submitting their bids.

You should inform bidders that it is their responsibility to check their bid before it is submitted to ensure accuracy and completeness.  Proportional due diligence should also be conducted prior to contract award. 

As always, you should aim to write your tender documents in a way that will result in clear, tangible, responses from your bidders.  This will mitigate against bidders trying to solely use AI generated responses, which tend to be generic and are unlikely to be sufficiently tailored to specific requirements outlined in the ITT.

What is Artificial Intelligence?

AI systems, tools and products are part of a rapidly growing and evolving market.  As the availability and functionality of AI tools grows, bidders may consider, or may already be using, AI (including generative AI – please see more detail on this below) to produce their bids.

Scotland’s AI Strategy sets out the vision for Scotland to become a leader in the development and use of trustworthy, ethical and inclusive AI.

The use of trustworthy, ethical and inclusive AI by bidders and buyers (in accordance with their public body’s policies and guidance) is not prohibited during the procurement process.  However, your organisation should undertake proportionate actions (in accordance with your policies and guidance) to mitigate any risks and maximise any benefits which AI may present.

Types of AI

There are several different types of AI systems that are increasingly used at present and offer a variety of applications and outputs.  You can find out more about these below:

The United Nations Convention on the Rights of the Child (Incorporation) (Scotland) Act 2024 (UNCRC Act) incorporates requirements set out under the UNCRC into domestic law, and makes provision to ensure compliance with duties relating to the UNCRC.

Section 6 of the UNCRC Act makes it unlawful for public bodies to act in a way that is incompatible with the UNCRC requirements when delivering relevant functions under powers conferred by an Act of the Scottish Parliament. This is referred to as the compatibility duty and applies to functions of a public nature. This includes functions of a public nature carried out under a contract with a public body. This duty comes into force on 16 July 2024.

A Child Rights and Wellbeing Impact Assessment (CRWIA) is a process which identifies any anticipated impact on children’s human rights and wellbeing as part of policy development. Where a CRWIA has been done, it and/or other relevant sources of information should inform the procurement process. Further information on CRWIA is available from the Scottish Government website. Although produced to assist Scottish Government officials comply with the Scottish Minister’s legal requirement to undertake CRWIAs, other organisations may find the guidance helpful.

Consideration of children’s rights and compatibility with UNCRC requirements will be especially relevant, for example, in a procurement to provide a direct service to children such as early learning, childcare and health services. However, there may be other services for which children will be an end user (alongside others) such as transportation, or services provided in public buildings and which could have an impact on access to children’s rights.

Further information on children’s rights is available from the Scottish Government website including non-statutory guidance to support public authorities and other organisations to deliver a children’s human rights approach in policy and practice.

There are no absolute criteria for defining the best route to market for innovation as every project is different, however in making your decision there are two important factors you should consider:  

• the complexity of your project (including the ability of your organisation to adopt the innovation), and
• the level of uncertainty, based on the nature and size of the research and development gap of the project

Level of Uncertainty (or the Research and Development Gap)

There is uncertainty in all innovation projects, whether it’s related to the outcome of the project, or parts of the project that are unknown.  However, as the project develops, you may realise there are metrics that an be recorded and measured over time.  

Unlike typical regulated procurement exercises, for innovation you may not have fully documented specific background information and/or a history of potential products or services.  This is particularly the case when the innovation project involves creating new goods or services that do not yet exist.

You may not for example have:

• product or service costs (either new or those from existing goods and services)
• supply chain partners in place or
• definitive solutions to solve the problem

Even if some research has been carried out in the Discovery and Definition phase, it is normal that levels of uncertainty will remain.  You may still be uncertain as to what the final solution will be, however you will have information gathered from your research to inform the decision for the most appropriate procurement procedure for your project. 

The further away from the market the potential goods or service is from implementation, the larger the R&D gap will be for the project.

Understanding the current Technology Readiness Level (TRL) of the goods or services helps to assess the gap between the current status of a product/service and the desired outcome.  

TRL’s help those involved in R&D to make decisions concerning the development and transitioning of technology. There are a number of online tools available to assist users in defining the TRL level such as those for Horizon Europe

Having defined the gap, you need to detail the R&D steps and resources required to close this gap.

The greater the uncertainty in closing the gap, the more likely it is that specific R&D is required to reduce that level of uncertainty. 

Complexity

Complexity refers to both organisational complexity and the technical complexity of the project.

It relates to areas such as the:

• project scope
• differences in users' needs
• organisational dynamics and structures
• policies and regulations affecting the project e.g. legal, data, systems integrations
• new technology complexities and
• skill sets of staff (both within the project and with entities outside of the project).

These factors will help you determine the type of process or procurement procedure that should be used. Please note the above is not a comprehensive list – every situation will have differing areas of complexity which need to be considered.

An example of differing complexity:

An innovation project within a single organisation is required to develop a new automated technology to receive payment for a visitor attraction.  This offers little complexity.

Whereas a project to create and codevelop highly specialised zero emissions vehicles, based on data from 3 pilot areas, for all public sector organisations, each with differing requirements, may present both technical and organisational complexities that require different consideration from a procurement procedure.

The below diagram shows how the level of complexity and uncertainty influences the procurement procedure chosen:

The above diagram is a matrix with two axis.  The x axis shows increasing complexity and the y axis shows increasing uncertainty due to an increasing research and development gap.  

A combination of the complexity and uncertainty axis result in four possible outcomes in terms of which procurement procedure to use.  These possible outcomes are:

Outcome 1: Low complexity and low R&D gap.  This can result in using either an open or restricted procurement procedure or the possible use of an existing Dynamic Purchasing System (DPS) or existing Framework Agreement.  The use of an existing DPS or Framework Agreement are subject to the applicability of the original terms of the Contract Notice.

Outcome 2: High complexity and low R&D gap.  This can result in using a procurement procedure involving negotiation where the highly complex aspects of the project would benefit from a process of negotiation working in partnership with contractors.  

As a result there are a number of negotiated procurement procedure options that can be used.  These include -

• Competitive Dialogue.  To use this procurement procedure the buyer may not have a precise understanding of the requirement and will use the competitive dialogue procedure to describe: the need and clarify/negotiate the solution.  However the buyer must set out the minimum requirements, award criteria and their weightings in the Contract Notices and procurement documents.  These cannot be changed during the negotiation process.  Competitive Dialogue allows bidders to submit initial solutions after being successful at the selection stage.  It allows buyers to negotiate proposed solutions with bidders.  Here the specification requirements concentrate on the buyer's needs without having to detail the nature, characteristics or solutions to be offered.
• Competitive Procedure with Negotiation This procedure lets you clarify bids with bidders after their submission of fully formed initial tenders.  You should use this procedure if you are unable to define how to meet your needs technically and/or you cannot specify the legal or financial requirements of your contract.
• Negotiated Procedure without Prior Publication.  This procedure should only be used in very exceptional circumstances where approval is given before any purchase is made.  The approver must not have been involved in the award of the contract.  Obligations under procurement legislation have been considered and the justification and approval must be formally recorded (along with robust supporting evidence) for audit purposes.
• Innovation Partnership.  This will be used in complex cases where there is no current solution available on the market to meet the need of the buyer.   Buyers can proceed with one or more partners undertaking separate R&D activities.  Innovation partnerships are structured in successive phases, based on the R&D steps undertaken e.g. design, manufacturing, etc., and you (as the buyer) may set intermediate targets that need to be met.

Outcome 3: Low complexity and high R&D gap.  This is where there is a large gap between what the market can supply and your needs. In such cases a research and development only phase may be used to determine whether and how the need can be met.  This research and development only phase (a Pre Procurement Process) can produce the necessary data to allow you to then carry out a further procurement exercise for implementation.  To aid your research, you should consider undertaking market analysis through a preliminary market consultation.

And finally Outcome 4: high complexity and high R&D gap.  This is is where a possible innovation project would benefit from trials or pilots, research and/or development to gauge its possible success or scalability.  In such situations it is usually thought that the project has potential and, as a result, other funding methods are sometimes used.  For example to conduct trials or a pilot you may have sourced finance via grants, endowment funds, charity funds, etc.  Where a pilot or trial has been carried out, for example via a pre-commercial procurement, a further procurement exercise is likely to be required to scale up the project.