Foundation session

Day 1
Understanding AI

By the end of today, AI will feel a lot less mysterious. We'll strip back the jargon, look at where it already shows up in your work, and practise using it safely and sensibly.

Make it clear

Understand AI, machine learning and generative AI in plain English.

Connect it to work

Spot where AI already shows up in the tools you use every day.

Practise with care

Try a generative AI tool using safe examples and sensible judgement.

01 / 33
Outcomes

By the end of today, you'll be able to...

1

Explain artificial intelligence in clear, everyday language.

2

Tell the difference between traditional software, machine learning and generative AI.

3

Recognise common examples of AI in workplace tools.

4

Describe, at a high level, how AI uses data to identify patterns.

5

Use a generative AI tool for a simple, low-risk task — and know exactly what to check.

02 / 33
How today works

This is a practical, supportive session

70%

Practice and examples

You will see short demonstrations, try tasks for yourself and apply the ideas to realistic scenarios.

20%

Conversation and comparison

You will learn from colleagues through paired discussion, group reflection and shared examples.

10%

Clear explanation

We'll introduce each idea briefly, then connect it straight to something you can use.

03 / 33
Starting point

Let's see what you already know

Open the first check in

Open the Day 1 resource and complete the opening check-in.

Purpose

This isn't about catching anyone out. It helps us see where you're already confident and where it's worth spending more time.

As you answer

Pay attention to why you choose an answer. Your reasoning matters more than the score.

04 / 33
Safe practice

Use fictional, anonymous or public examples only

Do not paste

Personal data, client records, employee records, private messages, contracts or restricted files.

Check first

Use approved tools, follow organisational policy, question the source and decide whether human review is needed.

Use instead

Fictional scenarios, public information, anonymous extracts and the practice text provided in the course.

05 / 33
Myth check

Some assumptions worth testing today

Assumption

AI is one single technology.

Assumption

AI understands things in the same way people do.

Assumption

AI always searches the internet before it answers.

Assumption

If the wording sounds confident, the answer must be correct.

06 / 33
Practice

How you'll practise today

One word AI opener

Share the first word AI brings to mind — then tell us what's behind it.

AI in my morning

Notice where AI might already be showing up in your day.

Tool taxonomy discussion

Connect different types of AI tools to the work you actually do.

First prompt practice

Try three safe prompts using an approved tool and non sensitive examples.

Opportunities and concerns matrix

Build a list of opportunities and concerns you'll carry through the whole course.

07 / 33
Activity rhythm

How each activity works

1

Understand the task

Check the purpose, timing, group setup and data boundary.

2

Try it

Work individually, with a partner or in a small group.

3

Compare

Notice what changed, improved, failed or surprised you.

4

Reflect

Connect the activity to your role, your judgement and your next sensible action.

08 / 33
Mental model

AI in plain English

AI is software that uses patterns in data to make predictions, recommendations or new content.

Keep this in mind whenever AI comes up. It makes the conversation much easier — and removes the mystery.

09 / 33
Analogy

Rulebook, learner, creator

Traditional software

People write the rules. The system follows those rules.

Machine learning

The system studies many examples and learns patterns from them.

Generative AI

The system creates text, images, audio or code based on patterns it has learned.

10 / 33
Everyday AI

AI may have helped before you logged on

Email

Spam filtering, priority inboxes and suggested replies.

Search

Ranking results, predicting search terms and recommending content.

Documents

Grammar suggestions, transcription and translation.

Systems

Scheduling, fraud detection and service routing.

11 / 33
A brief history

Three broad phases of AI

1950s to 1990s

Early AI focused on rules, expert systems and periods of slowed progress.

2000s to 2010s

More data and stronger computing helped machine learning become part of everyday technology.

2020s

Large language models and generative tools made AI more visible to the public.

Now

Workplaces are deciding where AI is useful, where it is risky and what good practice should look like.

12 / 33
How it learns

Training data, patterns and probability

Data

Examples of content or behaviour that the system can learn from.

Patterns

Relationships the system identifies across many examples during training.

Output

A likely response based on patterns, not a guarantee that the answer is true.

13 / 33
First practice

Your first safe prompt sequence

Summarise

Use an anonymous paragraph and ask for five clear key points.

Rewrite

Ask for a clearer version for someone who is not a specialist.

Draft

Create a fictional out of office message or a simple meeting note.

14 / 33
Activity timer

Focused practice sprint

15 minute activity timer

15:00

Your task

Work on your own or with your group. When we regroup, each person shares one thing that shifted their thinking.

15 / 33
As you work

What to notice and record

Notice

Look for patterns in your output, your group's output and the wider discussion.

Connect

Connect what you're seeing to the concept and the safe practice boundaries.

Improve

Make the next attempt stronger by adding context, checks or clearer instructions.

16 / 33
Decision practice

Use, use with caution, or do not use?

Draft a fictional meeting agenda for a team away day.

Discuss: use, use with caution, or do not use?

Paste employee sickness records into a public AI tool to summarise trends.

Discuss: use, use with caution, or do not use?

Use AI to draft client facing advice about a regulation you have not checked.

Discuss: use, use with caution, or do not use?

Ask AI to rewrite a generic internal paragraph in plainer English.

Discuss: use, use with caution, or do not use?

17 / 33
Reflection

Questions to help make sense of the activity

What worked better than you expected?

What felt unclear, unreliable or risky?

What would you need to check before using the output?

Where does human judgement still matter?

What would your workplace policy need to make clear?

What would you do differently next time?

18 / 33
Group discussion

What did we notice together?

Opportunities

Where could AI genuinely save time or improve quality?

Concerns

What felt risky, unclear or too confident?

Checks

What would need to be verified before workplace use?

Support

What guidance, policy or practice would help people use AI well?

19 / 33
Confidence check

Where are you now?

1 to 2

I would benefit from another example or more time to practise.

3

I understand the idea, but I would like support applying it.

4 to 5

I can try this independently and explain the checks I would make.

20 / 33
Evidence

What will you keep from today?

Evidence item

Your opportunities and concerns matrix, alongside a brief reflection on what changed in your thinking.

Quality test

It should show what you understood, what you tried and what judgement you applied.

21 / 33
Quality checkpoint

What thoughtful work looks like

DevelopingSecureStrong
Can repeat key terms, but still needs support to explain them clearly.Explains the concepts plainly and completes the guided tasks.Adapts the approach to context and identifies sensible checks.
Accepts or rejects AI without much explanation.Reviews the output and can name relevant risks.Balances usefulness, data limits, verification and accountability.
22 / 33
Knowledge check

Let's check what's landed

Open the quiz resource

Complete the Day 1 quiz and assessment resource.

Target

Aim for 7 or above. If a question stumps you, that's exactly what we want to talk about.

As you answer

Focus on the reasoning behind each answer, not just the final score.

23 / 33
Reflection

Before today I thought... now I think...

Before today

What assumption, worry or expectation did you bring into the session?

Now

What has changed in your understanding, confidence or judgement?

Write one honest sentence in your workbook — not what you think we want to hear.

24 / 33
Worked example

What is the AI doing in an email filter?

StepPlain English explanationWhat a person still checks
InputA message arrives with sender, wording, links, attachments and past patterns.Is the sender expected? Does anything feel unusual?
PredictionThe system compares the message with examples of spam, phishing and normal mail.A prediction is helpful, but not proof.
ActionThe email may be shown, hidden, flagged or moved to junk.Legitimate messages can be flagged and risky messages can slip through.

Why this example helps

It shows AI as pattern-based support, not a thinking person or a perfect judge.

25 / 33
Three terms

AI, machine learning and generative AI

TermSimple meaningExampleUseful caution
Artificial intelligenceA broad label for systems that perform tasks we associate with human judgement.A tool recommends which support ticket should be handled first.The label AI does not tell you whether the tool is good or safe.
Machine learningA system learns patterns from examples rather than only following hand-written rules.A fraud system notices unusual account behaviour.Patterns can reflect messy or biased data.
Generative AIA system creates new text, images, audio, code or plans from learned patterns.A chatbot drafts an email from bullet points.The result can sound polished and still be wrong.
26 / 33
Everyday examples

AI is often already in the workflow

Search

Ranking results, predicting search terms and suggesting related questions.

Documents

Grammar suggestions, transcription, translation and summarising.

Service

Routing queries, spotting common issues and prioritising cases.

Security

Spam filters, suspicious login alerts and fraud detection.

Discussion prompt

Which of these have you used without calling it AI? Which ones would you trust more or less, and why?

27 / 33
How it learns

From examples to a prediction

Example in training dataPattern the system may learnPossible output
Many urgent tickets mention outage, payroll, deadline or many affected users.Certain words and combinations often appear in urgent work.A new payroll deadline ticket is ranked as higher priority.
Many clear emails have short paragraphs, direct subject lines and next steps.Readers respond well to direct structure.A draft email includes a clear subject and action.
Many meeting summaries include decisions, actions and owners.These are common summary categories.The tool creates headings for decisions, actions and owners.

Important

This is not the same as understanding the organisation, the people or the consequences.

28 / 33
Generative AI

Why the output can feel convincing

It uses fluent language

The style can make the answer feel more certain than it is.

It fills gaps

If context is missing, the tool may still produce a complete-looking response.

It mirrors patterns

It can imitate common structures such as reports, policies and emails.

Useful phrase

A good-looking answer is a starting point. It is not evidence.

29 / 33
Safe prompt examples

Good first prompts use safe material

TaskSafe promptWhy it is safe
RewriteRewrite this fictional update so it is clearer for a busy colleague. Keep it under 120 words.Fictional text, clear task, easy to check.
SummariseSummarise this public article into five points for a non-specialist audience.Public source, named audience, visible output.
BrainstormSuggest five agenda items for a team meeting about communication, without using any real team names.No personal data and low stakes.
30 / 33
Risk changes by task

The same tool can be low risk or high risk

TaskLikely riskReason
Draft a fictional out of office message.LowNo personal data, easy to check, low consequence.
Summarise a confidential complaint.HighPersonal data, sensitive context and tone matter.
Suggest team meeting ideas.Low to mediumUseful starter, but still needs local judgement.
Rank job applicants.HighFairness, transparency and accountability issues.
31 / 33
Plain checks

Before trusting an AI answer, ask four questions

What is the source?

Is the answer based on text you supplied, public information or something unstated?

What could be wrong?

Look for dates, names, figures, assumptions and missing context.

Who is affected?

Would a mistake inconvenience, embarrass or disadvantage someone?

Who owns it?

If it is used at work, a person still owns the final output.

32 / 33
Close

Day 1 complete

You've got a clearer, more practical way to think and talk about AI at work — and the habit of checking before you trust.

Carry forward

Before you close: pick one thing you're taking forward and write it down.

33 / 33