Shadowing Practice: How AI Became More Expensive Than The Workers It Replaced - Learn English Speaking with Video
Elige un modo de práctica
Cada modo abre una sala de práctica a pantalla completa.
124 oraciones
Si las oraciones son demasiado cortas o largas, haz clic en Editar para ajustarlas.
1
In 2022, ChatGPT was released to the market.
2
And since then, the growth of AI has been impressive.
3
The creation of hyper -realistic images, code generation, and even the delegation of complex tasks.
4
AI seemed capable of doing everything with great ease.
5
And after seeing everything it was capable of, companies began adopting AI exponentially.
6
They incorporated it into their processes and realized that it was extremely cheap.
7
Bye!
8
Precisely because of its low costs, AI also began to bring problems.
9
"AI company called Anthropic is raising alarms tonight about AI's potential impact on employment that could soon be felt." Thousands of layoffs,
10
team reductions, and massive displacement.
11
Literally, AI was eliminating jobs across every industry.
12
from roles in tech companies, to fast food chains, to fast food chains, AI was wiping out almost any type of job.
13
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.
14
In recent months, a problem has started to emerge.
15
AI is becoming very expensive.
16
Because AI, maybe, is becoming too expensive to scale.
17
Using open claw these days is like driving a Ferrari.
18
So much so that Microsoft itself banned its own employees from using clawed code due to its high costs.
19
So Microsoft starts canceling clawed code licenses.
20
But this is not only happening at Microsoft.
21
Uber.
22
NVIDIA.
23
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.
24
Big tech is scrambling to slow down its use of AI.
25
Major companies have already exhausted their annual budgets And some say it's getting more difficult to justify the costs.
26
But if AI promised to be so cheap and efficient, How did it end up costing more than the workers it replaced?
27
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.
28
It promised efficiency and solutions to complex problems with just one click.
29
But most importantly, it promised to save companies a lot of money.
30
From technology companies to fast food chains, every business began implementing it at an unprecedented speed.
31
It was such a revolutionary technology that it could do the work of 10 people at one tenth of the price.
32
Or, at least, That was what people were saying.
33
"AI is a technology.
34
for taking away your job and maybe eventually displacing human sovereignty over our own future.
35
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.
36
And the way companies measured their AI performance and usage was through tokens.
37
Using a high number of tokens became somewhat of a status symbol among employees.
38
proving that they were going above and beyond to meet key performance indicators.
39
But there was a problem.
40
After realizing this, employees began using these tokens successively simply to appear more efficient at the end of the month.
41
This gave rise to what became known as token maxing.
42
"Token max?
43
Like, that's actually the coolest thing you can do now." "Token maxing
44
and this new way of building." But token maxing is expensive.
45
For text -based inquiries, the math is simple, with 750 words costing 1 ,000 tokens.
46
And generating videos, images, code, and more can cost much more.
47
By the end of 2025, token maxing began becoming popular among tech companies,
48
making it increasingly common for software engineers to use AI even for insignificant tasks.
49
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,
50
and that person said $5 ,000, I will go ape something else." Yes.
51
$250 ,000 worth of tokens, I am going to be deeply alarmed." Interestingly,
52
NVIDIA's CEO, one of the companies benefiting the most from AI adoption, strongly encouraged all developers to use these tokens.
53
The problem is that all this demand did not correspond to real demand but rather was artificially inflated.
54
Believe it or not, Silicon Valley engineers are now being scored on how many AI tokens they use.
55
The problem: that may end up inflating overall AI demand." At the same time,
56
Shortages of electronic components needed for building data centers are causing many of these projects to be delayed.
57
According to Bloomberg, nearly 50 % of all data center construction projects planned for 2026 in the US have been canceled
58
or delayed but many of these constructions and plans were based on the enormous demand that AI companies were receiving.
59
According to the Economic Times, demand for artificial intelligence infrastructure continues to greatly exceed supply.
60
The thing is
61
that experts are now beginning to question how much of this demand actually comes from real usage
62
and how much comes from inflated figures.
63
So when a hyperscaler like Andy Jassy at Amazon in his annual letter this morning
64
or that deal this morning tells Wall Street that demand is outstripping supply,
65
you gotta ask how much of that is real enterprise adoption and how much is engineers juicing a leaderboard?
66
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.
67
"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,
68
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.
69
In December of that year, the average cost was $1 .01 per million tokens,
70
and by May 2026, that number had reached $2 .12 per million tokens.
71
While this increase in price may seem insignificant, when taking into account the massive amount of tokens these companies use every day,
72
it translates into tens of millions of dollars per month.
73
use about $200 worth of tokens every week.
74
50 weeks a year, that's 10 ,000 bucks in tokens.
75
You have 40 ,000 employees, that's $400 million.
76
You have 90 ,000 employees, that's $900 million.
77
AT &T, for example, reported that it was consuming nearly 8 billion tokens every single day.
78
Without a doubt, this has been incredibly profitable for Anthropic and OpenAI, the leading AI companies in the world today.
79
According to Reuters, thanks to the adoption of AI agents at the enterprise level, Anthropic aims to nearly triple its annualized revenue in 2026.
80
To put that number into perspective, In 2025, Anthropic ended the year with revenue of around $3 billion,
81
and in 2026, that figure is expected to reach nearly $9 .6 billion.
82
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.
83
And AI companies are beginning to realize this.
84
The AI bill has come due.
85
Annual budgets blown through in weeks.
86
And now companies are facing a choice nobody has ever had to make before.
87
tokens or humans.
88
An internal memo from a Microsoft executive vice president instructed engineers to stop using Anthropics AI coding tools.
89
At first, Executives said it was to encourage the use of AI tools within the same ecosystem.
90
such as GitHub Copilot, But later it was discovered
91
that the main reason was the enormous costs generated by the use of Claude code.
92
And Microsoft is not alone in this, as Fortune reports that other companies are also pulling back on AI usage.
93
For example, Reuters reported that Meta consumed nearly 60 trillion tokens in a single month.
94
That's roughly 900 million dollars worth of Anthropix tokens based on the firm's blended prices.
95
If tech now costs what humans cost, every company in America is about to start asking which one to keep.
96
But these extra costs are becoming a real headache for corporate finance departments.
97
In fact, A Gartner survey found that three -quarters of the executives it polled expect technology budgets to increase this year,
98
and nearly half estimate double -digit percentage increases.
99
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.
Descargar app
Todo lo que necesitas para hablar con fluidez
Pronunciación con IAEvalúa cada oración
Práctica de IPADomina cada sonido
VocabularioConstruye tu banco de palabras
Juego de vocabularioAprende mientras juegas
Mismo canal
FOR YOU
Recomendar Video
B21
154 oraciones
The 7 Trillion AI Gamble Is Failing. Big Tech is TRAPPED Right NOW.
Business EnglishDaily ConversationUniversity Students
B24
57 oraciones
Labour force reskilling in AI age | Vietnam Today
Business EnglishUniversity StudentsWorking Adults
B22
25 oraciones
AI's Impact on Education: Pros, Cons, and Coping
Business EnglishUniversity StudentsWorking Adults
B217
130 oraciones
How AI Is Killing The Value Of A College Degree
Business EnglishInterview EnglishUniversity Students
B24
46 oraciones
Anthropic AI Researcher Issues Dire Warning it 'Could Kill Us All'
Business EnglishDaily ConversationUniversity Students
C1465
25 oraciones
The Human Economy in the Age of AI | Ben Pascut | TEDx
IELTS PrepBusiness EnglishUniversity Students
B21
34 oraciones
Top five moments from CNBC's interview with OpenAI CEO Sam Altman
Business EnglishUniversity StudentsWorking Adults
B22
391 oraciones
The Death of Educational Content on YouTube
Business EnglishUniversity StudentsWorking Adults
B23
63 oraciones
5 open source tools that replaced my $320/mo AI stack...
American AccentBusiness EnglishUniversity StudentsWorking Adults
B22
65 oraciones
Top 4 Humanoid Robots Set to Dominate 2026 | The Future of AI & Robotics
Business EnglishUniversity StudentsWorking Adults
B22
268 oraciones
Trump’s Canada Tariffs Are Backfiring in Red States | The Week
Business EnglishDaily ConversationUniversity Students
Vocabulario y notas de pronunciación para esta lección
Esta lección de conversación de nivel B2 se basa en el vídeo “How AI Became More Expensive Than The Workers It Replaced”. Las palabras que más se repiten: token, cost, companies, demand, tool. Este vídeo tiene 124 frases y 1741 palabras para practicar shadowing. La parte hablada dura 11:15. El hablante habla a un ritmo natural de unas 155 palabras por minuto, cercano a una conversación cotidiana. Solo el 80 % de las palabras está entre las 3.000 más comunes del inglés, por lo que el vocabulario es exigente.
Vocabulario clave de este vídeo
Las 15 palabras más avanzadas del vídeo, con su pronunciación y significado:
| Palabra | Pronunciación | Significado |
|---|---|---|
| token sustantivo | /ˈtoʊkən/ | seña |
| adoption sustantivo | /əˈdɑp.ʃən/ | adopción |
| inflate verbo | /ɪnˈfleɪt/ | inflar, hinchar |
| claw sustantivo | /klɔ/ | garra |
| exponentially adverbio | /ɛkspə(ʊ)ˈnɛnʃəli/ | exponencialmente |
| insignificant adjetivo | /ˌɪnsɪɡˈnɪfɪkənt/ | insignificante, oscuro |
| consume verbo | /kənˈsum/ | consumir |
| shortage sustantivo | /ˈʃɔː(ɹ)tɪd͡ʒ/ | falta, carestía |
| exceed verbo | /ɪkˈsiːd/ | exceder |
| profitable adjetivo | /ˈpɹɑ.fɪ.tə.bl̩/ | rentable, provechoso |
| justify verbo | /ˈd͡ʒʌstɪfaɪ/ | legitimar, justificar |
| cancel verbo | /ˈkæn.sl̩/ | tachar |
| artificially adverbio | artificialmente | |
| displace verbo | /dɪsˈpleɪs/ | desahuciar, desplazar |
| hyperscaler sustantivo | hiperescalador |
Phrasal verbs que vas a escuchar
| Palabra | Pronunciación | Significado |
|---|---|---|
| end up verbo | terminar, ir a parar | |
| lay off verbo | despedir, destituir | |
| sell off verbo | liquidar, rematar | |
| slow down verbo | desacelerar, decelerar | |
| take away verbo | /ˌteɪk əˈweɪ/ | quitar |
| wipe out verbo | aniquilar, exterminar |
Gramática en este vídeo
Las estructuras que más usa el hablante, con las palabras exactas del vídeo:
| Estructura | En el vídeo |
|---|---|
| Voz pasiva be + participio pasado — importa lo que ocurre, no quién lo hace | was released · be felt · was driven |
| Present perfect have/has + participio pasado — una acción pasada que sigue importando ahora | has been · has started · have reported |
Pronunciación a tener en cuenta
El hablante usa 3 contracciones y formas reducidas, como gonna, gotta, I'm. Dilas en su forma corta, tal como las oyes.
- Los sonidos de “th”: tenth /tɛnθ/, mathematics /mæθ(.ə)ˈmæt.ɪks/
- Los sonidos de “sh” y “zh”: adoption /əˈdɑp.ʃən/, exponentially /ɛkspə(ʊ)ˈnɛnʃəli/, shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, shareholder /ˈʃɛəˌhəʊl.də(ɹ)/, cancellation /kænsəˈleɪʃən/
- Palabras largas — cuida el acento: exponentially /ɛkspə(ʊ)ˈnɛnʃəli/, insignificant /ˌɪnsɪɡˈnɪfɪkənt/, cancellation /kænsəˈleɪʃən/, ecosystem /ˈikoʊˌsɪstəm/, unprecedented /ʌnˈpɹɛsɪdɛntɪd/
Sonidos difíciles para hispanohablantes:
- /v/ — no es /b/: los dientes tocan el labio inferior: sovereignty /ˈsɑv(ə)ɹənti/, inevitably /ɪˈnɛvɪtəbli/, investor /ɪnˈvɛs.tə(ɹ)/, revolutionary /ˌɹɛvəˈluʃənɛɹi/, developer /dɪˈvɛləpɚ/
- /z/ — sonora, no /s/: subsidize /ˈsʌbsɪdaɪz/, exhaust /ɪɡˈzɔst/, translate /tɹænzˈleɪt/
- /dʒ/ — no es la “y” ni la “ch”: shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, justify /ˈd͡ʒʌstɪfaɪ/, digit /ˈdɪd͡ʒɪt/, emerge /ɪˈmɝd͡ʒ/, generous /ˈd͡ʒɛn(ə)ɹəs/
Cómo practicar con este vídeo
- Escucha el vídeo entero una vez sin hablar y anota las palabras que no conoces.
- Empieza a velocidad 0,75×, haz shadowing frase por frase y vuelve a la velocidad normal cuando te resulte fácil.
- Grábate y compara con el original, prestando atención a palabras como token, adoption, inflate.
¿Qué es la Técnica de Shadowing?
Shadowing es una técnica de aprendizaje de idiomas respaldada por la ciencia, desarrollada originalmente para la formación de intérpretes profesionales y popularizada por el políglota Dr. Alexander Arguelles. El método es simple pero poderoso: escuchas audio en inglés nativo y lo repites en voz alta de inmediato, como una sombra que sigue al hablante con solo 1-2 segundos de retraso. A diferencia de la escucha pasiva o los ejercicios de gramática, el shadowing obliga a tu cerebro y músculos de la boca a procesar y reproducir simultáneamente patrones de habla reales. Las investigaciones muestran que mejora significativamente la precisión de la pronunciación, la entonación, el ritmo, el habla conectada, la comprensión auditiva y la fluidez al hablar, convirtiéndola en una de las metodologías más efectivas para la preparación del IELTS Speaking y la comunicación en inglés en el mundo real.