Artificial Intelligence:
From Technology to Practice

Real uses in our lives, our research, and our design

Mr. Husain Al-Saffar AI specialist and founder of Lumen Pro

The Big Picture

Not just chat

Artificial IntelligenceOne engine — many applications

Navigation & ETAFinding the best route and travel time

Recommendation systemsPicking what you see next from your behavior

Scientific image analysisSpotting details in medical and research images

Vehicle perceptionRecognizing what’s around the car, moment by moment

Image generationCreating a new image from a written description

Just the famous part Chatbots

Workshop roadmap

Our journey through 90 minutes

From the big idea to a hands-on project you build yourselves

We start here

1

AI around usTwo stories from everyday life

2

Generative AIAI that creates new content

3

Writing promptsCrafting a precise description the model understands

4

Image generationTurning words into images

5

DesignUsing results in polished visual work

6

Hands-on practiceA live walkthrough, step by step

Story 1 — Google Maps

You have a 4:00 appointment
at the research institute

It’s 3:35 — you open Maps and type the destination

ETA: 25 minutes — right on time

You head out, relaxed…

⌖ Research Center

25 m3:35 PM → 4:00 PM

Story 1 — Google Maps

Five minutes later… everything changed

3:35 You left — ETA 4:00

3:40 An accident reported on your route

3:40 Instant recalculation — alternate route

Estimated arrival ~ 25 minutes~ 33 minutes

How did Maps know about the accident
before you even got there?

Take a minute and guess: what information sources does the app rely on?

Story 1 — Google Maps

Thousands of signals were arriving in the background

⌖Research Center— 12 km

Live trafficWhat’s happening on the road right now

Historical traffic patternsHow this road usually behaves at this time every day

Road closuresAnnounced maintenance and detours

Nearby vehicle speedsDrivers’ phones send speed — without names

Thousands of anonymous signalsFrom other users’ phones on the same road

The system doesn’t just measure distance —it predicts from patterns

Story 2 — TikTok

Six moments… and the system watches them all

You open the app and start scrolling — without typing a single letter

1

Watched for two seconds then scrolled past

Weak signal

2

Skipped instantly

Not interested

3

Watched to the end

Interested

4

Rewatched twice

Very interested

5

Liked it

Clear feedback

6

Sent it to a friend

Strongest signal

▶

♥12K

✎340

↗Share

Discussion question

After these six signals…
what’s the next video?

Story 2 — TikTok

The answer: a loop that repeats with every scroll

1Show contentThe clip appears in front of the user

2User actionWatch, skip, like, share

3Update the behavior patternThe signal is added to the interest history

4New recommendationContent closest to your interests is recommended

↻ Continuous loop Each cycle makes recommendations sharper

The core idea

The system tracks patterns in your behavior
— it doesn’t read your thoughts

The feeling that “the app knows me” comes from millions of tiny signals you left yourself — not because it understands your personality

✓It notices what you do· ✓It compares your behavior to millions of users· ✗It doesn’t know what you’re thinking

Today we saw AI that predicts

But… can it create?

Yes —

Generative AI

GENERATIVE AI

Generative AI

This is the power of imagination

Ideas that didn’t exist… became images

A new idea

AI that doesn’t just predict — it produces

It learned from millions of examples… and now makes new ones that didn’t exist

TextArticles, messages, stories

ImagesDesigns, art, photoreal images

VideoMoving clips from a description

VoiceVoiceover in any tone and language

MusicFull tracks from words

CodePrograms and apps that actually run

Tools map

The popular tools — each with a specialty

The right tool depends on the kind of content you want

Text & chat

Writing, summarizing, and answering questions

ChatGPT

Claude

Gemini

Coding

Programs and apps that actually run

Cursor

GitHub Copilot

Replit

Audio & music

Voiceover, music, and audio summaries

ElevenLabs

Suno

NotebookLM

Tools map

Images & video — each with a specialty

The right tool depends on the visual content you want

Images

Designs, ads, and digital art

Nano Banana

GPT Image

Ideogram

SeeDream

Grok Imagine

QWEN

Wan

Video

Cinematic clips from text or an image

Veo

Kling

Seedance

Hailuo

Wan

Grok Imagine

Grok R2V

Edit & enhance

Edit an existing image or upscale its quality

Banana Edit

NanoBanana EditPro

SeeDream Edit

SeedEdit

QwenEdit

GPT Edit

Flux

Prunaai

Upscaler

All the image and video models — in one platform

+34 AI models, updated continuously

Subscribing to each tool alone is expensive and scattered… with Lumen Pro you get everything in one subscription — and more

ImagesGenerate · Edit · Enhance

★ Nano Banana ProNanoBanana 2SeeDream 5 ProSeedream 4.5Ideogram V4GPT Image 2.0Grok ImagineFluxQwen EditSeedEdit 4.5BananaEdit 2Upscaler★ Nano Banana ProNanoBanana 2SeeDream 5 ProSeedream 4.5Ideogram V4GPT Image 2.0Grok ImagineFluxQwen EditSeedEdit 4.5BananaEdit 2Upscaler

VideoFrom text or from an image

★ Veo 3.1★ Kling 2.6Seedance 2.0Seedance 1.5 ProHailuo 2.3Wan 2.6Grok ImagineGrok R2V★ Veo 3.1★ Kling 2.6Seedance 2.0Seedance 1.5 ProHailuo 2.3Wan 2.6Grok ImagineGrok R2V

The first Kuwaiti AI content platform at the regional level lumenpro.io

A question before the next chapter

So do all image tools
produce the same result?

Keep your answer in mind… we’ll discover it hands-on in a bit

Chapter 2

From words… to an image

How does AI understand what we say?

Can AI draw anything?

Yes.

But…

Live demo

Let’s try… now

✎

Cat

Generate ↵

Just one word — then we hit Enter

Is this the image you were expecting?

One word… isn’t enough. Let’s understand why.

Same request… two different results.

“A red sports car parked on a quiet city street, photorealistic, natural daylight.”

Image A

Image B

Same prompt, same model — but every generation is a new possibility.

What changed?

AI doesn’t see what’s in your head.

It builds the result from the information you give it.

Your imagination

The information you provide Like text, images, or any references

The AI Builds the result from the information it received

The result

5 examples…
And before each addition: what do you expect will change?

Prediction game — example 1 of 5

Arabic coffee pot

Arabic coffee pot

+Arabic coffee pot on a wooden table

+Morning light

+Kuwaiti diwaniya

+Ultra realistic

+85mm lens

+Steam

+Luxury photography

Arabic coffee pot on a wooden table, morning light

Arabic coffee pot on a wooden table in a Kuwaiti diwaniya, morning light

Arabic coffee pot on a wooden table in a Kuwaiti diwaniya, morning light, ultra realistic, 85mm lens, steam, luxury photography

Prediction game — example 2 of 5

Liberation Tower

Liberation Tower

+At night

+Neon lights

+Cyberpunk

+Rain

Liberation Tower at night, Neon lights, Cyberpunk

Prediction game — example 3 of 5

Old Kuwaiti house

Old Kuwaiti house

+Year 1950

+Rain

+Film photography

Old Kuwaiti house, 1950, rain, Film photography

Prediction game — example 4 of 5

Spain vs Argentina

Spain vs Argentina

+2026 World Cup final

+Inside a space stadium

+Epic cinematic

Spain vs Argentina, 2026 World Cup final

Spain vs Argentina, 2026 World Cup final, inside a space stadium

Spain vs Argentina, 2026 World Cup final, inside a space stadium, Epic cinematic

Prediction game — example 5 of 5

Lionel Messi — in Pixar style

Lionel Messi

+Pixar Style

!A real person — we need a reference photo

◆Real people need a reference photo + the person’s permission — text alone doesn’t know anyone’s face, and we don’t generate people’s images without consent

The takeaway

A strong prompt is built like Lego

Subject Place Lighting Camera Style Mood Colors Details

↓

One complete description — every piece narrows the possibilities and gets you closer to the image in your head

From pieces to a prompt

Each piece asks a question… the prompt answers it

SubjectWhat’s in the image? Who or what?

PlaceWhere is the scene happening?

LightingWhat kind of light? Morning, sunset, neon…

CameraAngle and lens? Close-up, wide, 85mm…

StylePhoto? Illustration? Pixar? Realistic?

MoodThe feeling: calm, dramatic, playful…

ColorsThe palette you want to see

DetailsThe small touches that make it yours

Full prompt: An Arabic coffee dallah on a wooden table inside a Kuwaiti diwaniya, soft morning light from the window, 85mm lens with shallow depth of field, luxury realistic photography, warm welcoming mood, earthy and golden tones, light steam rising from the cup

A question that leads us next

Two people write the exact same description …
They must get the same image, right?

Apparently… not always.

Answer to our earlier question

Same prompt — four different personalities

“A cat in a dishdasha drinking tea in a diwaniya” — literally the same sentence for all four

GPT Image — a cat in a dishdasha drinking tea in a diwaniya

GPT ImagePrecise, literal with details

Nano Banana 2 — a cat in a dishdasha drinking tea in a diwaniya

Nano Banana 2Google’s model — flexible and strong on detail

Flux — a cat in a dishdasha drinking tea in a diwaniya

FluxRealistic, natural, photo-like

Ideogram — a cat in a dishdasha drinking tea in a diwaniya

IdeogramBest at rendering text inside the image

◆So the prompt matters… andchoosing the right model matters just as much

A question that leads us next

If words have limits…
is there another way to get our intent across to AI?

Yes — we show it an image.

The pro secret

A reference image delivers what words can’t

Text only

“A golden falcon logo” ←

A nice result… but not what you imagined

Text + reference image

“A golden falcon logo” +

←

Closer result after adding the reference image

The model saw your intent — it didn’t just read it

The power of image reference

From a phone snap… to an ad shot

Same product — pro style from a single reference image

Before ↓

After

Before ↓

After

Before ↓

After

◆A phone photo becomes ad-ready — because the model saw the product, not just heard a description

The power of image reference

Logo stays… the scene changes

One Ashjar Cafe cup photo — same brand across three different settings

OriginalHandheld photo

Scene 1Green backdrop

Scene 2Palm shadow & sunlight

Scene 3Ad-style shot

◆The reference image locks the brand — logo, text, and art… even when everything else changes

Live example — Top Bun Restaurant

One photo… a full photoshoot

Same meal — new angles and setups from a single reference image

OriginalProduct photo

Shoot 1Close-ups and detail

Shoot 2Layouts and top-down

Shoot 3Motion and story in frame

◆The reference image keeps the product — AI changes the angle and composition

Handmade products

Before… and after

A simple product photo — becomes an ad shot

Before ↓

After

Before ↓

After

Before ↓

After

Before ↓

After

Before ↓

After

◆Same product… ad background and style — the reference image is what keeps the identity

Final challenge

Young Kuwaiti astronaut on dunes of a desert planet under two moons

If you were in my place…
what do you think was written in the Prompt?

SubjectA young Kuwaiti astronaut

Placestanding on dunes of a desert planet

Lightingunder two moons and cold starlight

Cameracinematic wide-lens shot

Styleultra-sharp photorealistic

Moodepic, inspiring atmosphere

A great image… isn’t luck. It’s a chain of decisions.

AI doesn’t read your thoughts…

it reads your words.

And every word changes the result.

We learned that every word changes the result.

Now… we apply it to the science poster’s image.

Chapter 3

AI improves your poster

Through the hero image

Real posters from KISR

This is what a science poster looks like

Food and Nutrition Group · Kuwait Institute for Scientific Research

Go for the Whole Grains Whole grains

The Effects of Lacking Macronutrients poster

Lacking Macronutrients Macronutrient balance

Go Healthy Choose healthier

◆The science and content… belong to the researchers. We focus on the image that stops the eye.

Our role today

We don’t write the poster… we design its image

1A science poster has content and layoutThat’s the researchers’ and designers’ work

2The hero image stops the eye in secondsThis is where AI comes in

3Our goal: a stronger image = a stronger posterWithout changing the science

◆A nice image… isn’t a poster by itself — but it can lift the whole poster

Where does AI help?

Three images that can upgrade your poster

Hero imageStops the eye the moment they see the poster

IllustrationExplains a science idea visually

Visual infographicTurns a number or comparison into an image

◆Today we focus on the hero image — the highest-impact slot

Same skill as Chapter 2

A weak prompt… and one that makes a poster image

Example: hero image for the “Whole Grains” poster

Weak “A healthy bread photo”

AI doesn’t know: which bread? what light? for which poster? for which audience?

A generic result… won’t stop the eye

Stronger ✓ A clear hero-image prompt

A sliced whole-wheat loaf with golden grains scattered around, soft window light, natural wood background, professional food photography, warm tones — suitable as the hero image for a science poster about whole grains

Every detail narrows the odds… and gets you closer to the image you want

◆The magic isn’t the tool… it’s how clearly you describe

Poster-image recipe

Three decisions… before you hit generate

1

The promptSubject + place + light + style + that it’s for a poster hero

2

The modelNot every model fits — pick one that matches the image type (food, illustration, realistic…)

3

The reference imageIf you have a composition or real product — a reference locks the result

◆Same Chapter 2 tools… aimed at the poster image

AI doesn’t design your poster…

you design its image.

It’s just a tool… in your hands.

The tool is in your hands…

the creativity is in you.

An initiative by STC

Best AI image challenge

Your turn… your image

Each group, from its own field in the KISR program, designs an AI image — with the prompt, model choice, and reference image.

There’s a prize for the best design.

A gift from us — free access

so you can finish what you started

KISR-2026

1 Scan the code 2 Sign up with your email 3 Enter the activation code

lumenpro.io

Try everything you learned today — free

If you forget everything today… remember this

The tool is in your hands… the creativity is in you.

Artificial Intelligence:From Technology to Practice

Not just chat

Our journey through 90 minutes

You have a 4:00 appointmentat the research institute

Five minutes later… everything changed

How did Maps know about the accidentbefore you even got there?

Thousands of signals were arriving in the background

Six moments… and the system watches them all

After these six signals…what’s the next video?

The answer: a loop that repeats with every scroll

The system tracks patterns in your behavior— it doesn’t read your thoughts

Today we saw AI that predicts

But… can it create?

This is the power of imagination

AI that doesn’t just predict — it produces

The popular tools — each with a specialty

Images & video — each with a specialty

All the image and video models — in one platform

So do all image toolsproduce the same result?

From words… to an image

Let’s try… now

Is this the image you were expecting?

Same request… two different results.

What changed?

AI doesn’t see what’s in your head.

5 examples…And before each addition: what do you expect will change?

Arabic coffee pot

Liberation Tower

Old Kuwaiti house

Spain vs Argentina

Lionel Messi — in Pixar style

A strong prompt is built like Lego

Each piece asks a question… the prompt answers it

Two people write the exact same description …They must get the same image, right?

Same prompt — four different personalities

If words have limits…is there another way to get our intent across to AI?

A reference image delivers what words can’t

From a phone snap… to an ad shot

Logo stays… the scene changes

One photo… a full photoshoot

Before… and after

If you were in my place…what do you think was written in the Prompt?

AI improves your poster

This is what a science poster looks like

We don’t write the poster… we design its image

Three images that can upgrade your poster

A weak prompt… and one that makes a poster image

Three decisions… before you hit generate

Your turn… your image

so you can finish what you started

Presenter shortcuts

Slides

Artificial Intelligence:
From Technology to Practice

You have a 4:00 appointment
at the research institute

How did Maps know about the accident
before you even got there?

After these six signals…
what’s the next video?

The system tracks patterns in your behavior
— it doesn’t read your thoughts

So do all image tools
produce the same result?

5 examples…
And before each addition: what do you expect will change?

Two people write the exact same description …
They must get the same image, right?

If words have limits…
is there another way to get our intent across to AI?

If you were in my place…
what do you think was written in the Prompt?