Pratique du Shadowing: AI Can Write Code. And That’s the Problem. - Apprendre l'anglais à l'oral avec la vidéo

Choisissez un mode d'entraînement

Chaque mode ouvre une salle d'entraînement en plein écran.

85 phrases

Si les phrases sont trop courtes ou trop longues, cliquez sur Edit pour les ajuster.

AI can write code now.

And not just a little, it's fast, it's convincing, and a lot of the time, it actually works.

You run it, and it does exactly what you expected.

And that's exactly where the problem begins.

Because in real-world engineering, working code is often the most dangerous kind of code.

Today, I want to break down why code that works can actually be a huge risk, especially when it's generated by AI, and I'll explain this from a real-world, production perspective.

Let's start with the first issue.

Bugs silently making it to production.

AI-generated code usually works for simple cases.

You test it quickly, it passes, and everything seems fine.

So what happens?

You trust it.

And your code review becomes less strict.

Now imagine this.

Edge cases are not handled.

Boundary conditions break the logic.

The system crashes when data grows.

These kinds of problems almost never show up in small tests.

If you test with 10 or 100 records, everything looks perfect. But production is different.

You have thousands, sometimes millions of records.

Unexpected inputs.

Real users doing unpredictable things.

And that's when things break.

Suddenly.

And you're left wondering, wait, this was working before.

But the truth is, it was always broken.

You just didn't hit the breaking point yet.

This happens all the time in real systems.

Second issue, design collapse.

AI is very good at local optimization.

It gives you code that works right now, for this specific moment.

But it does not think about your entire system architecture. So what happens?

The same logic gets duplicated in multiple places.

Naming becomes inconsistent.

Responsibilities become unclear.

At first, everything still works.

But later, when requirements change and they always do, things get messy.

You start asking questions like Where is this logic used?

Why do I have the same code in five different places?

Eventually, your system turns into spaghetti code And the worst part?

You reach a point where you're afraid to touch it It works, but any change might break something else That's technical debt Third issue,

security risks AI can generate insecure code without any warning.

For example, no protection against SQL injection, weak or missing authentication and authorization,

hardcoded API keys or secrets.

And again, this code works.

That's what makes it dangerous.

Because no one notices.

It passes tests.

It gets deployed.

And then one day, it becomes a real incident.

Data leaks Unauthorized access Working code caused a security problem That's not hypothetical,

it happens Fourth issue, performance problems AI does not consistently optimize for performance It might give you an O,

in superscript 2, solution without hesitation With small data, it runs instantly So again,

it looks fine But when your data scales, everything slows down dramatically.

You also see patterns like Unnecessary API calls Redundant database queries Individually, they seem harmless.

But together, they degrade your entire system.

And once users increase, your system starts to struggle.

This is a very common failure pattern.

When you rely on AI-generated code without truly understanding it.

You don't know why it works.

You can't fix it when it breaks.

You can't adapt it to new requirements.

And more importantly, you stop learning.

In the short term, you move faster.

But in the long term, you hit a wall.

Your growth as an engineer slows down or even stops.

So here's the conclusion.

AI is a tool that writes code.

But it is not an engineer that takes responsibility. Which means.

Working code does not mean correct code.

Working code might just mean.

You haven't hit the failure case yet.

So what actually matters going forward?

Your ability to evaluate code.

Can you explain why it works?

Can you analyze its time complexity?

Can you judge whether the design is good?

In the age of AI, writing code becomes easier.

But understanding code becomes more valuable.

The real skill is not writing.

It's seeing through the code.

That's all I have for you today.

If you enjoyed the video, please hit the like button and leave a comment.

See you in the next video.

Thank you.

Télécharger l'application

Tout pour parler couramment

Prononciation IAChaque phrase notée

Phonétique (IPA)Maîtrisez chaque son

VocabulaireEnrichissez votre lexique

Jeu de vocabulaireApprenez en jouant

Scanner le QR

Download on theApp Store Get it onGoogle Play

FOR YOU

Vidéo recommandée

195 phrases

Vocabulaire et conseils d’expression pour cette leçon

Cette leçon d’expression orale de niveau C1 s’appuie sur la vidéo « AI Can Write Code. And That’s the Problem ». Les mots qui reviennent le plus souvent : code, break, happens, test, issue. Cette vidéo contient 85 phrases et 691 mots à répéter en shadowing. La partie parlée dure 5:27. Le locuteur parle à un rythme régulier d’environ 127 mots par minute, confortable pour le shadowing. 85 % des mots font partie des 3 000 mots les plus courants en anglais ; le reste mérite d’être vu avant de commencer.

Vocabulaire clé de cette vidéo

Les 15 mots les plus avancés de la vidéo, avec leur prononciation et leur sens :

Mot	Prononciation	Sens
authentication nom	/ɔːˌθɛntɪˈkeɪʃn̩/	certification
optimize verbe	/ˈɑptɪmaɪz/	agir en optimiste
superscript nom		exposant, lettre supérieure
hesitation nom	/ˌhɛzɪˈteɪʃən/	hésitation
silently adverbe	/ˈsaɪləntli/	silencieusement
duplicate verbe	/ˈd(j)uː.plɪˌkeɪt/	copier, dupliquer
hypothetical adjectif	/ˌhaɪpəˈθɛtɪkəl/	hypothétique
query nom	/ˈkwɪɹ.i/	question
optimization nom	/ˌɑptəmaɪˈzeɪʃən/	optimisation
unpredictable adjectif		imprévisible
insecure adjectif	/ɪnsəˈkjʊɹ/	insécurisé
authorization nom	/ˌɔːθəɹaɪˈzeɪʃən/	autorisation
inconsistent adjectif	/ˌɪnkənˈsɪstənt/	incohérent, incompatible
harmless adjectif	/ˈhɑɹmləs/	inoffensif
analyze verbe	/ˈæn.əˌlaɪz/	analyser

Les verbes à particule que vous entendrez

Mot	Sens
slow down verbe	ralentir, décélérer
break down verbe	tomber en panne
see through verbe	voir à travers

Des phrases à répéter

Des phrases courtes et complètes de la vidéo, à réutiliser dans la conversation de tous les jours :

Let's start with the first issue.
You just didn't hit the breaking point yet.
Can you explain why it works?
Can you analyze its time complexity?
Can you judge whether the design is good?

Prononciation à surveiller

Le locuteur utilise 8 contractions et formes réduites, comme can't, you're, didn't. Prononcez-les sous leur forme courte, telles que vous les entendez.

Les sons « th »: authentication /ɔːˌθɛntɪˈkeɪʃn̩/, hypothetical /ˌhaɪpəˈθɛtɪkəl/, authorization /ˌɔːθəɹaɪˈzeɪʃən/
Les sons « sh » et « zh »: authentication /ɔːˌθɛntɪˈkeɪʃn̩/, hesitation /ˌhɛzɪˈteɪʃən/, optimization /ˌɑptəmaɪˈzeɪʃən/, authorization /ˌɔːθəɹaɪˈzeɪʃən/, injection /ɪnˈd͡ʒɛk.ʃən/
Mots longs — placez bien l’accent: authentication /ɔːˌθɛntɪˈkeɪʃn̩/, hesitation /ˌhɛzɪˈteɪʃən/, hypothetical /ˌhaɪpəˈθɛtɪkəl/, optimization /ˌɑptəmaɪˈzeɪʃən/, authorization /ˌɔːθəɹaɪˈzeɪʃən/

Les sons difficiles pour les francophones :

/h/ — il se souffle, il n’est pas muet: hesitation /ˌhɛzɪˈteɪʃən/, hypothetical /ˌhaɪpəˈθɛtɪkəl/, harmless /ˈhɑɹmləs/
/r/ anglais — langue recourbée, sans frotter la gorge: degrade /dɪˈɡɹeɪd/, query /ˈkwɪɹ.i/, insecure /ɪnsəˈkjʊɹ/, authorization /ˌɔːθəɹaɪˈzeɪʃən/, harmless /ˈhɑɹmləs/

Comment s’entraîner avec cette vidéo

Écoutez la vidéo en entier une fois sans parler et notez les mots que vous ne connaissez pas.
Répétez phrase par phrase à vitesse normale, en reprenant chacune jusqu’à ce que votre rythme corresponde à celui du locuteur.
Enregistrez-vous et comparez avec l’original, en faisant attention à des mots comme authentication, optimize, superscript.

Qu'est-ce que la technique du Shadowing ?

Le Shadowing est une technique d'apprentissage des langues fondée sur la science, développée à l'origine pour la formation des interprètes professionnels. Le principe est simple mais puissant : vous écoutez de l'anglais natif et le répétez immédiatement à voix haute — comme une ombre suivant le locuteur avec un décalage de 1 à 2 secondes. Les recherches montrent une amélioration significative de la précision de la prononciation, de l'intonation, du rythme, des liaisons, de la compréhension orale et de la fluidité.

Technique du shadowing : lire le guide complet étape par étape →

Pratique du Shadowing: AI Can Write Code. And That’s the Problem. - Apprendre l'anglais à l'oral avec la vidéo

Choisissez un mode d'entraînement

Télécharger l'application

AI Destroyed My Profession

STOP Running Code On Your Computer

900+ hours of Learning AI Agents in 9 Minutes

Are we heading to an era of stagnation or innovation?

25 crazy software bugs explained

GitHub's AI Problem - How AI Is Damaging the Platform

Generative AI explained in 2 minutes

Advantages and Disadvantages of AI

How the Algorithm Hijacked Monkey’s Brain (Why Monkey Can’t Learn Online)

Jack Dorsey’s New Dessa AI Voice Sounds Scary Human

Why Software Factories Fail

How I’d Learn Coding in 2026 (Without Wasting Years)