Module 5 — Ethics, AI Safety & Your Future

🎬

1. The State of AI Ethics Right Now

Watch: the real debates happening in AI — jobs, bias, safety, power and who decides

+40 XP

01

Video not loading? Watch directly on YouTube → Dec 2025 · TEDx Youth · English

// Three things to watch for

→ How the speaker — a student himself — frames AI's impact on young people specifically

→ His argument about who controls AI — and why that matters for everyone

→ What ethical standards he says we need — and whether you agree

Rate this lesson:

saved ✓

🎭

2. Can You Tell What's Real? — The Deepfake Question

The question everyone should be asking. Here's the honest answer — and the data is alarming

+70 XP

02

// YOUR QUESTION — MODULE 4

"In the near future, would an AI video of a human actually not be any different to a real life human — and would we still be able to tell the difference?"

That is one of the most important questions anyone can ask about AI in 2026. Here's the honest, research-backed answer.

// THE NUMBERS — 2025/2026 DATA

8M

Deepfakes projected to be shared in 2025 — up from 500,000 in 2023

Europol, 2025

90%

Of all online content may be AI-generated by 2026, Europol estimates

European Parliament, 2025

9%

Of adults feel confident they can identify deepfakes — only 1 in 11

European Parliament, 2025

// THE HONEST ANSWER TO YOUR QUESTION

Within 2–3 years: for most people, most of the time — no. We won't be able to tell.

The technology is improving faster than our ability to detect it. AI-generated video of humans already has near-perfect lip sync, accurate facial micro-expressions, and emotionally convincing voice modulation. The tells that existed two years ago — blurry edges, unnatural blinking, background glitches — are rapidly disappearing.

Researchers describe this as an arms race: detection AI improves, then generation AI leaps ahead again. Right now, generation is winning. The most concerning development in 2026 is that audio deepfakes are already effectively undetectable to the human ear — meaning a phone call that sounds exactly like your parent or a friend could be entirely AI-generated.

The solution the research community is converging on is not better human detection — it's digital watermarking: embedding invisible signals into AI-generated content so tools can verify authenticity. But this only works if everyone adopts it. Open-source tools and bad actors can bypass it easily.

// SPOT THE TELL — CAN YOU IDENTIFY DEEPFAKES FROM DESCRIPTIONS?

For each description below, decide: is this more likely real footage or AI-generated? Then see how the technology actually behaves.

// WHAT TO LOOK FOR — CURRENT TELLS (changing fast)

👁️

Blinking patterns

AI-generated faces historically blink less than humans, or blink in unnatural rhythms. Getting harder to spot as models improve.

👂

Ear and hair edges

Complex geometry — hair strands, earrings, detailed edges — still trips up generation models. Look for smearing or blurring at the margins.

🦷

Teeth and hands

Hands are notoriously difficult for AI — wrong number of fingers, unnatural poses. Teeth can look unnaturally perfect or slightly off.

💡

Lighting consistency

Light source on the face should match the environment. AI sometimes gets this wrong — shadows fall in the wrong direction.

🎤

Audio-visual sync

Lip movements not quite matching speech, especially on complex sounds. Modern deepfakes have mostly solved this — but not perfectly.

🧠

Your best defence

Pause before reacting. Check the source. Search for corroboration. Emotional content designed to shock is the primary delivery mechanism for deepfakes.

Rate this lesson:

saved ✓

⚖️

3. Ethics Debate Lab — Pick a Side, Defend It

Four real AI dilemmas. You pick a side. AI argues back. You have to rebut it

+60 XP

03

These aren't hypothetical debates — they're happening right now in parliaments, boardrooms, and courtrooms. Pick a dilemma, choose your position, read the strongest counterargument, then write your rebuttal. There's no right answer — the skill is in the quality of your reasoning.

Should AI make hiring decisions?

Companies are increasingly using AI to screen CVs, conduct initial interviews, and score candidates — sometimes without any human reviewing rejected applications. Supporters say AI removes human bias. Critics say AI embeds it. Some firms have cut their HR teams by 60% using these systems.

POSITION A

AI hiring is fairer — it removes human prejudice and judges candidates on merit alone

POSITION B

AI hiring is dangerous — it embeds historical bias and removes human judgement from decisions that change lives

// STRONGEST COUNTERARGUMENT

Now write your rebuttal — why does your position still hold despite that counterargument?

Should police use facial recognition AI?

UK police have been trialling AI facial recognition cameras in public spaces — matching faces against databases of suspects in real time. In 2024, the Metropolitan Police made over 100 arrests using the technology. But multiple innocent people have been wrongly detained. The system is significantly less accurate on darker skin tones.

POSITION A

Facial recognition makes communities safer — catching criminals faster and deterring crime in public spaces

POSITION B

Facial recognition is a dangerous surveillance tool — disproportionately harmful to minorities and a threat to civil liberties

// STRONGEST COUNTERARGUMENT

Now write your rebuttal — why does your position still hold?

Is AI-generated art theft from human artists?

AI image generators like Midjourney are trained on billions of images scraped from the internet — including the work of living artists — without permission or payment. The AI then generates images "in the style of" those artists, potentially replacing the work they built careers on. Several major lawsuits are ongoing in the US and UK.

POSITION A

AI art is a new creative tool — just like cameras or Photoshop, and artists should adapt rather than resist technology

POSITION B

AI art companies are profiting from stolen work — artists deserve compensation and the right to opt out of training data

// STRONGEST COUNTERARGUMENT

Now write your rebuttal — why does your position still hold?

Should AI be allowed to make lethal military decisions?

Several countries — including the US, China, Israel and Russia — are developing autonomous weapons systems that can identify and engage targets without human approval. Supporters argue AI makes faster, more precise decisions than humans under fire. Critics call for an international ban, arguing that no machine should ever be authorised to take a human life.

POSITION A

Autonomous weapons can be more precise and faster than humans, potentially reducing civilian casualties in conflict

POSITION B

No AI should ever be authorised to take a human life — this decision must always remain with a human being

// STRONGEST COUNTERARGUMENT

Now write your rebuttal — why does your position still hold?

Rate this lesson:

saved ✓

🚀

4. Your Future — The Jobs That Will Actually Exist

The roles paying top salaries in 2030 — and how everything you've learned connects to them

+60 XP

04

Companies are paying professionals with AI skills 56% more than those without. These aren't distant future jobs — they're being hired for right now, and they'll be mainstream by the time you enter the workforce. Here are the roles that matter most.

🎯

Context & Prompt Engineer

£45,000–£120,000

Designing the information environments and instruction systems that make AI perform at its best. You started learning this in Module 2 and went deeper in Module 3.

🤝

Human-AI Collaboration Designer

£55,000–£130,000

Deciding what AI does and what humans do in any given workflow. You designed workflows in Module 4 — this is that skill professionalised.

🔍

AI Auditor & Ethics Officer

£50,000–£140,000

Checking AI systems for bias, errors, hallucination and ethical issues before they cause harm. 60% of enterprises will have ethics boards by end of 2026.

⚙️

AI Automation Architect

£60,000–£150,000

Building the workflow systems that connect AI to business processes at scale. The automation you built in Module 4 is a micro version of what these architects do daily.

🛡️

AI Security Specialist

£65,000–£160,000

Protecting AI systems from attack — prompt injection, data poisoning, adversarial inputs. Gartner forecasts $2.5 trillion in AI cybersecurity spend in 2026 alone.

🧠

Chief AI Officer (CAIO)

£150,000–£400,000+

Leading an organisation's entire AI strategy. 1 in 4 companies now has one. 66% expect most companies to hire one within two years. The ceiling of this career path.

// 10% AHEAD — MULTI-AGENT SYSTEMS (coming 2027-2028)

The next leap: AI as a coordinated team, not a single tool

Right now, most people use one AI tool at a time. What's coming — already being deployed at enterprise level — is multi-agent systems: teams of specialised AI agents working together, each handling a different part of a complex task, coordinated by a human director.

Imagine: you need to launch a product. One AI agent researches the market. Another writes the copy. Another designs the assets. Another builds the landing page. Another analyses the results. All coordinated by you — the human who sets the goal, makes the judgement calls, and owns the outcome.

// The role that emerges: Agent Orchestrator

→ Defines the goal and success criteria

→ Designs which agents do which tasks

→ Monitors quality at each handoff point

→ Makes the judgement calls AI can't

→ This is a human role. It will be highly paid. You're being prepared for it right now.

// HOW THIS COURSE CONNECTS TO YOUR FUTURE

MOD 1

Mental models + vocabulary — Every job in AI requires you to speak the language. You built the foundation.

MOD 2

Prompt + context engineering — A skill already paying £45-120k. You started building it early.

MOD 3

Tool literacy + critical evaluation — Knowing which tool to use and when to not trust it is the core of AI auditing.

MOD 4

Workflow design + automation — The foundation of Human-AI Collaboration Design and Automation Architecture.

MOD 5

Ethics + critical thinking — AI Ethics Officer is one of the fastest-growing roles in tech. It starts here.

Rate this lesson:

saved ✓

🎓

5. Capstone Project — Design Your Own AI Product

The Level 1 finale. Invent something real. This is your portfolio piece

+70 XP

05

You've completed four modules. You understand AI, you can direct it, you know the tools, you've built workflows, and you understand the ethics. Now use everything — design an original AI product. Not a copy of something that exists. Something you'd actually want.

STEP 1 — THE PROBLEM

What problem does your product solve?

The best products start with a genuine problem someone actually has — not a solution looking for a use case. What frustrates you, or people you know, that AI could genuinely help with?

STEP 2 — THE SOLUTION

What does your AI product do?

Describe how it works — what the user does, what the AI does, and what the output is. Keep it simple: "The user does X. The AI does Y. The user gets Z."

STEP 3 — WHO IS IT FOR?

Describe your ideal user

Age, situation, what they care about. The more specific you are, the better the product. "Everyone" is not a user.

STEP 4 — THE AI QUESTION

What exactly does the AI do in your product — and what does the human always control?

Using what you learned in Module 4 — draw the human-AI boundary. What's automated, what needs human judgement?

STEP 5 — THE ETHICS CHECK

What could go wrong — and how have you designed against it?

Every AI product has potential harms. What are yours, and how does your design reduce them?

Fill in all 5 sections to complete your capstone

Rate this lesson:

saved ✓

Ethics, Safety
& Your Future

Your Learning Path

Ethics, Safety& Your Future

Your Learning Path

Ethics, Safety
& Your Future