Pratique du Shadowing: How AI Became More Expensive Than The Workers It Replaced - Apprendre l'anglais à l'oral avec la vidéo

Choisissez un mode d'entraînement

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

124 phrases

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

In 2022, ChatGPT was released to the market.

And since then, the growth of AI has been impressive.

The creation of hyper -realistic images, code generation, and even the delegation of complex tasks.

AI seemed capable of doing everything with great ease.

And after seeing everything it was capable of, companies began adopting AI exponentially.

They incorporated it into their processes and realized that it was extremely cheap.

Bye!

Precisely because of its low costs, AI also began to bring problems.

"AI company called Anthropic is raising alarms tonight about AI's potential impact on employment that could soon be felt." Thousands of layoffs,

team reductions, and massive displacement.

Literally, AI was eliminating jobs across every industry.

from roles in tech companies, to fast food chains, to fast food chains, AI was wiping out almost any type of job.

It could perform the work of entire teams in just minutes, largely because it was much more efficient, but above all because it was cheaper to maintain.

In recent months, a problem has started to emerge.

AI is becoming very expensive.

Because AI, maybe, is becoming too expensive to scale.

Using open claw these days is like driving a Ferrari.

So much so that Microsoft itself banned its own employees from using clawed code due to its high costs.

So Microsoft starts canceling clawed code licenses.

But this is not only happening at Microsoft.

Uber.

NVIDIA.

Amazon and hundreds of other companies have reported similar cost problems, making it almost impossible even for top executives to justify such a high expense.

Big tech is scrambling to slow down its use of AI.

Major companies have already exhausted their annual budgets And some say it's getting more difficult to justify the costs.

But if AI promised to be so cheap and efficient, How did it end up costing more than the workers it replaced?

Although the birth of artificial intelligence dates back to the beginning of the 1950s, it was not until 2022 that this technology truly experienced a boom.

It promised efficiency and solutions to complex problems with just one click.

But most importantly, it promised to save companies a lot of money.

From technology companies to fast food chains, every business began implementing it at an unprecedented speed.

It was such a revolutionary technology that it could do the work of 10 people at one tenth of the price.

Or, at least, That was what people were saying.

"AI is a technology.

for taking away your job and maybe eventually displacing human sovereignty over our own future.

Starting in 2023, companies began not only encouraging, but also forcing their employees to use AI tools, causing demand for these models to grow uncontrollably.

And the way companies measured their AI performance and usage was through tokens.

Using a high number of tokens became somewhat of a status symbol among employees.

proving that they were going above and beyond to meet key performance indicators.

But there was a problem.

After realizing this, employees began using these tokens successively simply to appear more efficient at the end of the month.

This gave rise to what became known as token maxing.

"Token max?

Like, that's actually the coolest thing you can do now." "Token maxing

and this new way of building." But token maxing is expensive.

For text -based inquiries, the math is simple, with 750 words costing 1 ,000 tokens.

And generating videos, images, code, and more can cost much more.

By the end of 2025, token maxing began becoming popular among tech companies,

making it increasingly common for software engineers to use AI even for insignificant tasks.

But much of this popularity was driven by statements like these: "That $500 ,000 engineer, at the end of the year, I'm gonna ask him how much did you spend in tokens,

and that person said $5 ,000, I will go ape something else." Yes.

$250 ,000 worth of tokens, I am going to be deeply alarmed." Interestingly,

NVIDIA's CEO, one of the companies benefiting the most from AI adoption, strongly encouraged all developers to use these tokens.

The problem is that all this demand did not correspond to real demand but rather was artificially inflated.

Believe it or not, Silicon Valley engineers are now being scored on how many AI tokens they use.

The problem: that may end up inflating overall AI demand." At the same time,

Shortages of electronic components needed for building data centers are causing many of these projects to be delayed.

According to Bloomberg, nearly 50 % of all data center construction projects planned for 2026 in the US have been canceled

or delayed but many of these constructions and plans were based on the enormous demand that AI companies were receiving.

According to the Economic Times, demand for artificial intelligence infrastructure continues to greatly exceed supply.

The thing is

that experts are now beginning to question how much of this demand actually comes from real usage

and how much comes from inflated figures.

So when a hyperscaler like Andy Jassy at Amazon in his annual letter this morning

or that deal this morning tells Wall Street that demand is outstripping supply,

you gotta ask how much of that is real enterprise adoption and how much is engineers juicing a leaderboard?

But due to token maxing, the cancellation of data center construction projects, and the massive adoption of AI worldwide, the cost of those tokens has increased significantly.

"Aught of concerns over the rising cost of artificial intelligence, it led to a major tech stock sell -off that dragged down the S &P and NASDAQ,

which fell more than 500 points." According to Bloomberg Finance, The average cost of tokens from large language models, or LLMs, has more than doubled since 2025.

In December of that year, the average cost was $1 .01 per million tokens,

and by May 2026, that number had reached $2 .12 per million tokens.

While this increase in price may seem insignificant, when taking into account the massive amount of tokens these companies use every day,

it translates into tens of millions of dollars per month.

use about $200 worth of tokens every week.

50 weeks a year, that's 10 ,000 bucks in tokens.

You have 40 ,000 employees, that's $400 million.

You have 90 ,000 employees, that's $900 million.

AT &T, for example, reported that it was consuming nearly 8 billion tokens every single day.

Without a doubt, this has been incredibly profitable for Anthropic and OpenAI, the leading AI companies in the world today.

According to Reuters, thanks to the adoption of AI agents at the enterprise level, Anthropic aims to nearly triple its annualized revenue in 2026.

To put that number into perspective, In 2025, Anthropic ended the year with revenue of around $3 billion,

and in 2026, that figure is expected to reach nearly $9 .6 billion.

But as a result, employers are frequently blowing through their AI budgets, putting them in a situation where it's no longer clear that using AI is cheaper than hiring people.

And AI companies are beginning to realize this.

The AI bill has come due.

Annual budgets blown through in weeks.

And now companies are facing a choice nobody has ever had to make before.

tokens or humans.

An internal memo from a Microsoft executive vice president instructed engineers to stop using Anthropics AI coding tools.

At first, Executives said it was to encourage the use of AI tools within the same ecosystem.

such as GitHub Copilot, But later it was discovered

that the main reason was the enormous costs generated by the use of Claude code.

And Microsoft is not alone in this, as Fortune reports that other companies are also pulling back on AI usage.

For example, Reuters reported that Meta consumed nearly 60 trillion tokens in a single month.

That's roughly 900 million dollars worth of Anthropix tokens based on the firm's blended prices.

If tech now costs what humans cost, every company in America is about to start asking which one to keep.

But these extra costs are becoming a real headache for corporate finance departments.

In fact, A Gartner survey found that three -quarters of the executives it polled expect technology budgets to increase this year,

and nearly half estimate double -digit percentage increases.

A sizeable chunk of that spending will go toward AI, which is expected to account for more than one -fifth of total enterprise technology spending by 2035.

100

And the numbers prove it.

101

Global spending on AI agents and models is rising at concerning rates.

102

By 2027, spending on this technology is estimated to reach $680 billion, more than double what companies spent in 2025.

103

But these estimates were made using the current average costs of these tools, and considering that both and OpenAI are still not profitable.

104

they are operating at a loss.

105

But that is about to change because in the coming months, both companies are expected to go public and will have to answer to shareholders.

106

Therefore, they will inevitably increase the prices of using tokens from their AI tools and agents in order to generate profits.

107

In other words, the amount of money companies are currently paying for tokens will multiply exponentially.

108

So, if the cost of using these tools continues to rise, of tokens.

109

The thing is, This is not a prediction.

110

In fact, this is already happening.

111

Maybe human labor will be more cost efficient after all.

112

Uber and Microsoft are starting to second -guess aggressive AI adoption.

113

That's partly because the cost of using AI can exceed the cost of human employees.

114

According to Reuters, the current cost of running AI agents is already comparable to the cost of human labor

115

and in some cases, it is already more expensive.

116

While in coding tasks, the use of AI is undeniably cheaper.

117

In data entry tasks, the difference is minimal.

118

But in others, such as call centers, the cost of human labor is actually cheaper.

119

Although AI companies have so far been able to subsidize the real cost of tokens due to generous and abundant funding, that appears to be coming to an end.

120

According to the Wall Street Journal, enterprise customers are beginning to question the true return on these investments.

121

And the reality is that with the public listings of Anthropic and OpenAI, combined with investor pressure to see returns and electronic component shortages,

122

this creates the perfect formula for token costs to continue increasing.

123

And if maintaining these models is already expensive today, There will come a point when companies start asking themselves whether it makes sense to pay more for a simple tool.

124

than what they used to pay many of the workers they laid off.

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

Même chaîne

Why Even Harvard’s Smartest Graduates Can’t Get a Job Now

Daily ConversationUniversity StudentsWorking Adults

Why Companies Are Quietly Rehiring Software Engineers

Business EnglishUniversity StudentsWorking Adults

FOR YOU

Vidéo recommandée

154 phrases

Vocabulaire et conseils d’expression pour cette leçon

Cette leçon d’expression orale de niveau B2 s’appuie sur la vidéo « How AI Became More Expensive Than The Workers It Replaced ». Les mots qui reviennent le plus souvent : token, cost, companies, demand, tool. Cette vidéo contient 124 phrases et 1741 mots à répéter en shadowing. La partie parlée dure 11:15. Le locuteur parle à un rythme naturel d’environ 155 mots par minute, proche d’une conversation courante. Seuls 80 % des mots font partie des 3 000 mots les plus courants en anglais, le vocabulaire est donc exigeant.

Vocabulaire clé de cette vidéo

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

Mot	Prononciation	Sens
token nom	/ˈtoʊkən/	symbole
adoption nom	/əˈdɑp.ʃən/	adoption
inflate verbe	/ɪnˈfleɪt/	gonfler, se gonfler
claw nom	/klɔ/	griffe
insignificant adjectif	/ˌɪnsɪɡˈnɪfɪkənt/	insignifiant
consume verbe	/kənˈsum/	consommer
shortage nom	/ˈʃɔː(ɹ)tɪd͡ʒ/	défaut, rareté
exceed verbe	/ɪkˈsiːd/	excéder
profitable adjectif	/ˈpɹɑ.fɪ.tə.bl̩/	profitable, fructueux
justify verbe	/ˈd͡ʒʌstɪfaɪ/	justifier
cancel verbe	/ˈkæn.sl̩/	annuler, résilier
artificially adverbe		artificiellement
instruct verbe	/ɪnˈstɹʌkt/	instruire, enseigner
layoff nom		chômage technique, licenciement économique
leaderboard nom		classement

Les verbes à particule que vous entendrez

Mot	Prononciation	Sens
end up verbe		se terminer, se finir
lay off verbe		virer, licencier
sell off verbe		vendre à perte, brader
slow down verbe		ralentir, décélérer
take away verbe	/ˌteɪk əˈweɪ/	retirer
wipe out verbe		anéantir

La grammaire de cette vidéo

Les structures que le locuteur utilise le plus, avec les mots exacts de la vidéo :

Structure	Dans la vidéo
Voix passive be + participe passé — l’accent est mis sur ce qui arrive, pas sur qui le fait	was released · be felt · was driven
Present perfect have/has + participe passé — une action passée qui compte encore maintenant	has been · has started · have reported

Prononciation à surveiller

Le locuteur utilise 3 contractions et formes réduites, comme gonna, gotta, I'm. Prononcez-les sous leur forme courte, telles que vous les entendez.

Les sons « th »: tenth /tɛnθ/, mathematics /mæθ(.ə)ˈmæt.ɪks/
Les sons « sh » et « zh »: adoption /əˈdɑp.ʃən/, exponentially /ɛkspə(ʊ)ˈnɛnʃəli/, shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, shareholder /ˈʃɛəˌhəʊl.də(ɹ)/, cancellation /kænsəˈleɪʃən/
Mots longs — placez bien l’accent: exponentially /ɛkspə(ʊ)ˈnɛnʃəli/, insignificant /ˌɪnsɪɡˈnɪfɪkənt/, cancellation /kænsəˈleɪʃən/, ecosystem /ˈikoʊˌsɪstəm/, unprecedented /ʌnˈpɹɛsɪdɛntɪd/

Les sons difficiles pour les francophones :

/h/ — il se souffle, il n’est pas muet: shareholder /ˈʃɛəˌhəʊl.də(ɹ)/, headache /ˈhɛdeɪk/
/tʃ/ et /dʒ/ — à ne pas adoucir en « ch » et « j »: shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, justify /ˈd͡ʒʌstɪfaɪ/, chunk /t͡ʃʌŋk/, digit /ˈdɪd͡ʒɪt/, emerge /ɪˈmɝd͡ʒ/
/r/ anglais — langue recourbée, sans frotter la gorge: shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, uber /ˈuːbəɹ/, profitable /ˈpɹɑ.fɪ.tə.bl̩/, instruct /ɪnˈstɹʌkt/, outstrip /ˌaʊtˈstɹɪp/

Comment s’entraîner avec cette vidéo

Écoutez la vidéo en entier une fois sans parler et notez les mots que vous ne connaissez pas.
Commencez à la vitesse 0,75×, répétez phrase par phrase, puis revenez à la vitesse normale quand cela devient facile.
Enregistrez-vous et comparez avec l’original, en faisant attention à des mots comme token, adoption, inflate.

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: How AI Became More Expensive Than The Workers It Replaced - Apprendre l'anglais à l'oral avec la vidéo

Choisissez un mode d'entraînement

Télécharger l'application

Why Even Harvard’s Smartest Graduates Can’t Get a Job Now

Why Companies Are Quietly Rehiring Software Engineers

The 7 Trillion AI Gamble Is Failing. Big Tech is TRAPPED Right NOW.

Labour force reskilling in AI age | Vietnam Today

AI's Impact on Education: Pros, Cons, and Coping

How AI Is Killing The Value Of A College Degree

Anthropic AI Researcher Issues Dire Warning it 'Could Kill Us All'

The Human Economy in the Age of AI | Ben Pascut | TEDx

Top five moments from CNBC's interview with OpenAI CEO Sam Altman

The Death of Educational Content on YouTube

5 open source tools that replaced my $320/mo AI stack...

Top 4 Humanoid Robots Set to Dominate 2026 | The Future of AI & Robotics

Trump’s Canada Tariffs Are Backfiring in Red States | The Week