meet users’ needs, consider if:

• there’s data containing the

information you need, even if

disguised or buried

• it’s ethical and safe to use the

data - refer to the Data Ethics

Framework4

• you have a the right sort of data

for the AI model to learn from

• the task is large scale and

repetitive enough that a human

would struggle to carry it out

• it would provide information a

team could use to achieve

outcomes in the real world

useful to consider whether:

• the models are achieving their

purpose and business objectives

• there is a clear accountability

framework for models in

production

• there is a clear testing and

monitoring framework in place

• your team has reviewed and

validated the code

• the algorithms are robust,

unbiased, fair and explainable

• the project fits with how citizens

and users expect their data to

be used

Depending on your organisation’s

maturity, it may be useful to set up

a dedicated board, committee or

forum to handle AI training data

and model governance.

It can be useful to keep a central

record of all AI technologies you

use, listing:

• where an AI model is in use

• what the AI model is used for

• who’s involved

• how it’s assessed or checked

• what other teams rely on the

technology

make sure you’re hypothesis-led

and can constantly iterate to best

help your users and their needs.

You should integrate your AI

systems development with your

wider project phases.

1. Discovery - consider your

current data state, decide

whether to build, buy or

collaborate, allocate

responsibility for AI models,

assess your existing data, build

your AI team, get your data

ready for AI, and plan your AI

modelling phase.

2. Alpha - build and evaluate your

machine learning model.

3. Beta - deploy and maintain your

model.

You should consider AI ethics and

safety throughout all phases.

Significant time is needed to

understand the feasibility of using

your data in a new way. This means

the discovery phase tends to be

longer and more expensive than

for services without AI.

Your data scientists may be familiar

with a lifecycle called CRISP-DM7

and may wish to integrate parts of

it into your project.

Discovery can help you understand

the problem that needs to be

solved.

sources

You should:

• thoroughly understand the

problem and the needs of

different users

• assess whether an AI system is

the right tool to address the

user needs

• understand the processes and

how the AI model will connect

with the wider service

• consider the location and

condition of the data you will

use

should assess your existing data.

Training an AI system on errorstrewn

data can result in poor

results due to:

• the dataset not containing clear

patterns for the model to

explore when making a

prediction

• the dataset containing clear but

accidental patterns, resulting in

the model learning biases

You can use a combination of

accuracy, completeness,

uniqueness, timeliness, validity,

relevancy, representativeness,

sufficiency or consistency to see if

the data is high enough quality for

an AI system to make predictions

from.8

When assessing your AI data, it’s

useful to collaborate with someone

who has deep knowledge of your

data, such as a data scientist. They

will be familiar with the best

practice for measuring, cleaning

and maintaining good data

standards for ongoing projects.

Make your data proportionate to

user needs and understand the

limitations of the data to help you

assess your data readiness.

• do you understand the onward

effects of using data in this way?
is the data accurate and

complete and how frequently is

the data updated?

• is the data representative of the

users the model’s results will

impact?

• was the data gathered using

suitable, reliable, and impartial

sources of measurement?

• is the data secure and do you

have permission to use it?

• what modelling approaches

could be suitable for the data

available?

• do you have access to the data

and how quickly can you access

it?

• where is the data located?

• what format is the data in and

does it require significant

preparation to be ready for

modelling?

• is your data structured - for

example, can you store it in a

table, or unstructured such as

emails or webpages?
are there any constraints on the

data - for example, does it

contain sensitive information

such as home addresses?

• can you link key variables within

and between datasets?

• If you’re unsure about your use

of data, consult the Data Ethics

Framework guidance9 to check

your project is a safe application

and deployment of AI models.

should be multidisciplinary, with a

diverse combination of roles and

skills to reduce bias and make sure

your results are as accurate as

possible. When working with AI you

may need specialist roles such as a:

• data architect to set the vision

for the organisation’s use of

data, through data design, to

meet business needs

• data scientist to identify complex

business problems while

leveraging data value - often

having at least two data

scientists working on a project

allows them to better

collaborate and validate AI

experiments

delivery of data products and

services into systems and

business processes

• ethicist to provide ethical

judgements and assessments on

the AI model’s inputs

• domain expert who knows the

environment where you will be

deploying the AI model results -

for example, if the model will be

investigating social care,

collaborate with a social worker

implementation, you should

identify how you can best integrate

AI with your existing technology

and services.

It’s useful to consider how you’ll

manage:

• data collection pipelines to

support reliable model

performance and a clean input

for modelling, such as batch

upload or continuous upload

• storing your data in databases

and how the type of database

you choose will change

depending on the complexity of

the project and the different

data sources required

• data mining and data analysis of

the results

• any platforms your team will use

to collate the technology used

across the AI project to help

speed up AI deployment

When choosing your AI tools, you

should bring in specialists, such as

data scientists or technical

architects to assess what tools you

currently have to support AI.
Use Cloud First when setting up

your infrastructure.10

software tool which helps teams

connect all of the technology they

require across their project

workflow, speeding up AI

deployment and increasing the

transparency and oversight over AI

models.

When deciding on whether to use a data science platform, it’s useful to consider how the platform can:

• provide access to flexible

computation which allows teams

to have secure access to the

power needed to process large

amounts of data

• help your team build workflows

for accessing and preparing

datasets and allow for easy

maintenance of the data

• provide common environments

for sharing data and code so the

team can work collaboratively

• let your teams clearly share their

output through dashboards and

applications
provide a reproducible

environment for your teams to

work from

• help control and monitor

project-specific or sensitive

permissions

data quality, you should prepare

your data to make sure it is secure

and unbiased. You may find it

useful to create a data factsheet

during discovery to keep a record

of your data quality.

In the same way you should have

diversity in your team, your data

should also be diverse and

reflective of the population you are

trying to model. This will reduce

conscious or unconscious bias.

Alongside this, a lack of diverse

input could mean certain groups

are disadvantaged, as the AI model

may not cater for a diverse set of

needs. You should read the Data

Ethics Framework guidance to

understand the limitations of your

data and how to recognise any bias

present.

You should also:

• evaluate the accuracy of your

data, how it was collected, and

consider alternative sources

• consider the social context of

where, when and how the

system is being deployed

• consider if any particular groups

might be at an advantage or

disadvantage in the context in

which the system is being

deployed

Keeping your data secure

Make sure you design your system

to keep data secure. To help keep

data safe:

• follow the National Cyber Security

Centre’s guidance

(www.ncsc.gov.uk) on using data

with AI models

• make sure your system is

compliant with GDPR and DPA

201811

As with any other software, you

should design and build modular,

loosely coupled systems which can

be easily iterated and adapted.

Most of the data in government

available to train our models is

within legacy systems which might

contain bias and might have poor

controls around it. For legacy

systems to be compatible with AI

technology, you will often need to

invest a lot of work to bring your

legacy systems up to modern

standards.

You’ll also need to carefully

consider the ethical and legal

implications of working with historic

data and whether you need to seek

permission to use this information.

Evaluate your data preparation

phase

When you complete your data

preparation phase you should

have:

• a dataset ready for modelling in

a technical environment
a set of features (measurable

properties) generated from the

raw dataset

• a data quality assessment using

a combination of accuracy, bias,

completeness, uniqueness,

timeliness/currency, validity or

consistency

Researching the end to end

service

During the discovery phase, you

should explore the needs of the

users of the end to end service.

Like other digital services, you’ll use

this phase to determine whether

there’s a viable service you could

build that would solve user needs,

and that it’s cost-effective to

pursue the problem.

You’ll be able to check guidance on

how to know when your discovery

is finished before moving on to

alpha.

Plan and prototype your AI

model build and service

If you have decided to build your AI

model in-house, you should follow

these steps.

1. Split the data.

2. Create a baseline model.

3. Build a prototype of the model

and service.

4. Test the model and service.

5. Evaluate the model.

6. Assess and refine performance.

Split the data

Your team will need to train the

models they build on data. Your

team should split your data into a:

• training set to train algorithms

during the modelling phase

• validation set for assessing the

performance of your models

• test set for a final check on the

performance of your best model

Create a baseline model

Your team should build a simple

baseline version model before they

build any more complex models.

This provides a benchmark that

your team can later compare more

complex models against, and will

help your team identify problems in

your data.

Build a prototype of the model

and service

Once you have a baseline model,

your team can start prototyping

more complex models. This is a

highly iterative process, requiring

substantial amounts of data, and

will see your team probably build a

number of AI models before

deciding on the most effective and

appropriate algorithm for your

problem.

Keeping your team’s first model

simple and setting up the right

end-to-end infrastructure will help

smooth the transition from alpha

to beta. You can action this by

focusing on the infrastructure

requirements for your AI pipelines

as the same time your team is

developing the model. Your simple

model will provide baseline metrics

and information on the model’s

behaviour that you can use to test

more complex models.

Test the model and service

Your team will need to test your

models throughout the process to

mitigate against issues such as

overfitting or underfitting that

could undermine your model’s

effectiveness once deployed.

Your team should only use the test

set on your best model. Keep this

data separate from your models

until this final test. This test will

provide you with the most accurate

impression of how your model will

perform once deployed.

Evaluate the model

Your team will need to evaluate

your model to assess how it is

performing against unseen data.

This will give you an indication of

how your model will perform in the

real world.

The best evaluation metric will

depend on the problem you are

trying to solve, and your chosen

model. While you should select the

evaluation metric with data

scientists, you should also consider

the ethical, economical and societal

implications. These considerations

make the fine tuning of AI systems

relevant to both data scientists and delivery leads.

Choose the final model

When choosing your final model,

you will need to consider:

• what level of performance your

problem needs

• how interpretable you need

your model to be

• how frequently you need

predictions or retraining

• the cost of maintaining the

model

Once you select a final model, your

team will need to assess its

performance, and refine it to make

sure it performs as well as you

need it to. When assessing your

model’s performance consider:

• how it performs compared to

simpler models

• what level of performance you

need before deploying the

model

• what level of performance you

can justify to the public, your

stakeholders, and regulators

• what level of performance

similar applications deliver in

other organisations

• whether the model shows any

signs of bias

If a model does not outperform

human performance, it still may be

useful. For example, a text

classification algorithm might not

be as accurate as a human when

classifying documents, however

they can perform at a far higher

scale and speed than a human.

Evaluate your Alpha phase

When you complete building your

AI prototyping phase, you should

have:

• a final model or set of predictive

models and a summary of their

performance and characteristics

• a decision on whether or not to

progress to the beta phase

• a plan for your beta phase

integrating the model into the

service’s decision-making process

and using live data for the model to

make predictions on.

Using your model in your service

has three stages.

performance-test the model

with live data and integrate it

within the decision-making

workflow. Integration can

happen in a number of ways,

from a local deployment to the

creation of a custom application

for staff or customers. This

decision is dependent on your

infrastructure and user

requirements.

undertake continuous

evaluation to make sure the

model still meets business

objectives and the model is

performing at the level required.

This will make sure the model

performance is in line with the

modelling phase and to help you

identify when to retrain the

model.

users feel confident in using,

interpreting, and challenging any

outputs or insights generated by

the model.

You should continue to collect user

needs so your team can use the

model’s outputs in the real world.

When moving from alpha to beta,

there are some best-practice

guidelines to smooth the transition.

After creating a beta version, you

team can use automated testing to

create some high-level tests before

moving to more thorough testing.

Working in this way means you can

launch new improvements without

worrying about functionality once

deployed.

During alpha, you will have relied

mostly on data scientists to assess

the opportunity and your data

state.

with a strong knowledge of devops,

servers, networking, data

stores, data management, data

governance, containers, cloud

infrastructure and security design.

This skillset is likely to be better

suited to an engineer rather than a

data scientist so maintaining a

cross-functional team will help

smooth the transition from alpha

to beta.

When you complete your beta

phase, you should have:

• AI running on top of your data,

learning and improving its

performance, and informing

decisions

• a monitoring framework to

evaluate the model’s

performance and rapidly identify

incidents

• launched a private beta followed

by a public end-to-end beta

prototype which users can use

in full

• found a way to measure your

service’s success using new data

you’ve got during the beta phase
evidence that your service

meets government

accessibility requirements

• tested the way you’ve

designed assisted digital

support for your service

to make sure the model shows no

signs of bias or discrimination. You

can consider whether:

• the algorithm is performing in

line with safety and ethical

considerations

• the model is explainable

• there is an agreed definition of

fairness implemented in the

model

• the data use aligns with the Data

Ethics Framework

• the algorithm’s use of data

complies with privacy and data

processing legislation

the model is achieving its purpose/

business objectives. You can

consider whether:

• the model solves the problem

identified

• how and when you will evaluate

the model

• the user experience aligns with

existing government guidance

provides a clear accountability

framework for the model. You can

consider:

• whether there is a clear and

accountable owner of the model

• who will maintain the model

• who has the ability to change

and modify the code

monitoring makes sure a robust

testing framework is in place. You

can consider:

• how you will monitor the

model’s performance

• who will monitor the model’s

performance

• how often you will assess the

mode

protects against reputational risks

arising from the application of the

model. You can consider whether:

• the project fits with the

government organisation’s use

of AI systems

• the model fits with the

government organisation’s

policy on data use

• the project fits with how citizens/

users expect their data to be

used

consider whether:

• the team has validated the code

• the code is open source

substantial impact for individuals,

communities, and society. To make

sure the impact of your AI project is

positive and does not

unintentionally harm those affected

by it, you and your team should

make considerations of AI ethics

and safety a high priority.

This section introduces AI ethics

and provides a high-level overview

of the ethical building blocks

needed for the responsible delivery

of an AI project.

The following guidance is designed

to complement and supplement

the Data Ethics Framework. The

Framework is a tool that should be

used in any project.12

Ethical considerations will arise at

every stage of your AI project. You

should use the expertise and active

cooperation of all your team

members to address them.

Understanding what AI

ethics is

AI ethics is a set of values,

principles, and techniques that

employ widely accepted standards

to guide moral conduct in the

development and use of AI

systems.

The field of AI ethics emerged from

the need to address the individual

and societal harms AI systems

might cause. These harms rarely

arise as a result of a deliberate

choice - most AI developers do not

want to build biased or

discriminatory applications or

applications which invade users’

privacy.

cause involuntary harm are:

• misuse - systems are used for

purposes other than those for

which they were designed and

intended

• questionable design - creators

have not thoroughly considered

technical issues related to

algorithmic bias and safety risks

• unintended negative

consequences - creators have

not thoroughly considered the

potential negative impacts their

systems may have on the

individuals and communities

they affect

The field of AI ethics mitigates

these harms by providing project

teams with the values, principles,

and techniques needed to produce

ethical, fair, and safe AI

applications.

projects using AI

The guidance summarised in this

chapter and presented at length in

The Alan Turing Institute’s further

guidance on AI ethics and safety is

as comprehensive as possible.

However, not all issues discussed

will apply equally to each project

using AI.

An AI model which filters out spam

emails, for example, will present

fewer ethical challenges than one

which identifies vulnerable

children. You and your team should

formulate governance procedures

and protocols for each project

using AI, following a careful

evaluation of social and ethical

impacts.

Establish ethical building

blocks for your AI project

You should establish ethical

building blocks for the responsible

delivery of your AI project. This

involves building a culture of

responsible innovation as well as a

governance architecture to bring

the values and principles of ethical,

fair, and safe AI to life.

responsible innovation

To build and maintain a culture of

responsibility you and your team

should prioritise four goals as you

design, develop, and deploy your AI

project. In particular, you should

make sure your AI project is:

• ethically permissible - consider

the impacts it may have on the

wellbeing of affected

stakeholders and communities

• fair and non-discriminatory -

consider its potential to have

discriminatory effects on

individuals and social groups,

mitigate biases which may

influence your model’s outcome,

and be aware of fairness issues

throughout the design and

implementation lifecycle

• worthy of public trust -

guarantee as much as possible

the safety, accuracy, reliability,

security, and robustness of its

product

• justifiable - prioritise the

transparency of how you design

and implement your model, and

the justification and

interpretability of its decisions

and behaviours
Prioritising these goals will help

build a culture of responsible

innovation. To make sure they are

fully incorporated into your project

you should establish a governance

architecture consisting of a:

• framework of ethical values

• set of actionable principles

• process based governance

framework

These values:

• provide you with an accessible

framework to enable you and

your team members to explore

and discuss the ethical aspects

of AI

• establish well-defined criteria

which allow you and your team

to evaluate the ethical

permissibility of your AI project

You can read further guidance on

SUM Values in The Alan Turing

Institute’s comprehensive guidance

on AI ethics and safety.

ethical values

You should understand the

framework of ethical values which

support, underwrite, and motivate

the responsible design and use of

AI. The Alan Turing Institute calls

these ‘the SUM Values’:

• respect the dignity of individuals

• connect with each other

sincerely, openly, and inclusively

• care for the wellbeing of all

• protect the priorities of social

values, justice, and public

interest

principles

While the SUM Values can help you

consider the ethical permissibility

of your AI project, they are not

specifically catered to the

particularities of designing,

developing, and implementing an

AI system.

AI systems increasingly perform

tasks previously done by humans.

For example, AI systems can screen

CVs as part of a recruitment

process. However, unlike human

recruiters, you cannot hold an AI

system directly responsible or

accountable for denying applicants

a job.


This lack of accountability of the AI

system itself creates a need for a

set of actionable principles tailored

to the design and use of AI

systems. The Alan Turing Institute

calls these the ‘FAST Track

Principles’:

• fairness

• accountability

• sustainability

• transparency

Carefully reviewing the FAST Track

Principles helps you:

• ensure your project is fair and

prevent bias or discrimination

• safeguard public trust in your

project’s capacity to deliver safe

and reliable AI

If your AI system processes social

or demographic data, you should

design it to meet a minimum level

of discriminatory non-harm. To do

this you should:

• use only fair and equitable

datasets (data fairness)
include reasonable features,

processes, and analytical

structures in your model

architecture (design fairness)

• prevent the system from having

any discriminatory impact

(outcome fairness)

• implement the system in an

unbiased way (implementation

fairness)

Accountability

You should design your AI system

to be fully answerable and

auditable. To do this you should:

• establish a continuous chain of

responsibility for all roles

involved in the design and

implementation lifecycle of the

project

• implement activity monitoring to

allow for oversight and review

throughout the entire project

Sustainability

The technical sustainability of these

systems ultimately depends on

their safety, including their

accuracy, reliability, security, and

robustness.

You should make sure designers

and users remain aware of:

• the transformative effects AI

systems can have on individuals

and society

• your AI system’s real-world

impact

Transparency

Designers and implementers of AI

systems should be able to:

• explain to affected stakeholders

how and why a model performed

the way it did in a specific

context

• justify the ethical permissibility,

the discriminatory non-harm,

and the public trustworthiness of

its outcome and of the

processes behind its design and

use

To assess these criteria in depth,

you should consult The Alan Turing

Institute’s guidance on AI ethics and

safety.

governance framework

The final method to make sure you

use AI ethically, fairly, and safely is

building a process-based

governance framework. The Alan

Turing Institute calls it a ‘PBG

Framework’. Its primary purpose is

to integrate the SUM Values and

the FAST Track Principles across

the implementation of AI models

within a service.

Building a good PBG Framework for

your AI project will provide your

team with an overview of:

• the relevant team members and

roles involved in each

governance action

• the relevant stages of the

workflow in which intervention

and targeted consideration are

necessary to meet governance

goals

• explicit time frames for any

evaluations, follow-up actions,

re-assessments, and continuous

monitoring

• clear and well-defined protocols

for logging activity and for

implementing mechanisms to

support end-to-end auditability