Peru · AI and machine learning · Ages 13 to 18

AI course for teens in Peru

A teenager in Peru can use an AI app in seconds. Knowing whether its answer is any good takes a course. This page is for secundaria students and their parents who want the deeper version: writing Python, training small models on real or invented data, and, above all, testing them so the score means something. It covers what such a course should include, what school in Peru already gives, which AI tools a 13 to 17 year old may legally use, and a lesson about a model that cheats without meaning to.

Book the free first class Book a priority demo Try the leakage lesson Ask on WhatsApp

Live classes since 2020 · 10,000+ students worldwide · taught live, online and in English

The short version

Modern Age Coders runs an AI course for teens in Peru: live online classes, taught in English, for students aged 13 to 18. Teenagers learn Python, work with data, train small machine learning models and learn to test them honestly, so they know when a model is good and when it only looks good. AI tools are used only where the vendor's age terms and a parent's permission allow. A mentor based in India teaches a group of fifteen to twenty teenagers or one student alone, at flexible hours on Peru time.

Choose a starting course

Three AI courses for teenagers

Pick by what the student already knows. None of the three needs earlier coding experience, but the second and third suit a student who enjoys maths.

AI Tools Course for Teens (13+) course thumbnail

AIT / 01

AI Tools Course for Teens

For 13 to 17 year olds who want to use AI well before building it: prompting, checking answers and real projects, only with tools whose terms allow the student's age and with a parent's permission where required.

Open the syllabus →
AI and Machine Learning for Teens course thumbnail

AIT / 02

AI and Machine Learning for Teens

The core course. Python from the beginning, then data, models and evaluation, including the leakage lesson below. School maths is enough to start.

Open the syllabus →
Data Science for Teens course thumbnail

AIT / 03

Data Science for Teens

For students drawn to numbers and questions: cleaning data, charts, and first machine learning models in Python, with a strong habit of checking every result.

Open the syllabus →

What teens learn

What should a teenager learn in an AI course?

Three things, in this order: enough Python to build a model, enough understanding of data to know what the model learned from, and enough discipline in testing to know whether it works. Most short courses teach only the first.

Build

Write Python that loads data, trains a simple model and makes predictions. Students start with models small enough to understand by hand, such as a single threshold, before they meet libraries that hide the arithmetic.

Understand the data

Where did the examples come from, who labelled them, what is missing? A model can only repeat the patterns in its data, so a student who cannot describe the data cannot explain the model.

Test honestly

Keep some data back, never let it leak into training, and report the result on examples the model has never seen. This is the habit that separates a project that works from one that only scores well.

Google's free Machine Learning Crash Course, written for developers, puts the third point bluntly: "The only fair test of a model is against new examples, not duplicates." It also warns that test sets "wear out" when the same one is used again and again to make decisions. We teach those two sentences to teenagers early, because almost every surprising result in a student project comes from breaking one of them.

By the end of the core course a student should be able to take a small dataset, train two different models, compare them fairly, and write a short report in English explaining which is better and why they believe it. That report, more than any certificate, shows what they learned.

Secundaria and AI

Does secundaria in Peru teach AI or programming?

Not as a timetabled subject. The national curriculum, the CNEB, has no computing area in secundaria; digital skills are a transversal competency spread across other subjects. MINEDU has, however, started several projects around programming and AI.

Recent MINEDU programming and AI initiatives for school students
InitiativeWhenWhat MINEDU announced
Fábrica de ProgramadoresAnnounced 24 July 2025Python training for 1,000 students in 4th and 5th of secundaria in Lima Norte, with the option of the PCEP certification exam
Hackathon 2025Results published 4 August 2025Fourth edition, with categories for basic-education students and for teachers; most projects used AI and a STEAM approach
Girls taking part2023 to 2025Female student participation in the Hackathon grew 28% across the three editions

Sources: MINEDU press notes of 24 July 2025 (Fábrica de Programadores) and 4 August 2025 (Hackathon 2025), gob.pe; CNEB plan of studies. Read on 9 November 2026. We are not connected with MINEDU or either initiative.

What this means for a student

A pilot for 1,000 students in one part of Lima is a real start, and its choice of Python matches what we teach. But most secundaria students in Peru will finish 5th year without a single timetabled lesson on how a program or a model works.

INEI's household survey shows the audience is ready: 92.5% of Peruvians aged 12 to 18 used the internet in the second quarter of 2026, up from 89.7% in 2023. Almost every teenager is online; very few are taught what is happening inside the apps they use.

Where an outside course fits

An AI course works well alongside school, not instead of it. School maths (percentages, graphs, a little algebra) is exactly what a first machine learning course needs, and teenagers often find that maths makes more sense once they use it to judge a model.

For students in Spanish-medium schools, the English is a second benefit; for those at bilingual or IB schools, it is simply the language they already study in. Our Spanish-language page curso de IA para adolescentes explains the same courses for parents who prefer to read in Spanish.

Tools at 13 to 17

Which AI tools can a 13 to 17 year old use?

Some, with conditions. Each company sets its own minimum age in its terms, and several require a parent's permission until 18. We follow those terms exactly, and we would rather a student learned less about one tool than broke its rules.

What the vendors' own terms say about teenagers, as we read them
ToolMinimum age in the termsHow we use it with 13 to 17 year olds
ChatGPT (OpenAI)13; under 18 needs a parent or guardian's permissionOnly with a parent's permission, to question and check work
Gemini (Google)13 with a personal or school accountOccasionally, for comparison with other tools
GitHub Copilot13Late in a course, after the student can write the code alone
Claude (Anthropic)18Not used by anyone under 18
Python, scikit-learn, notebooksNo AI account neededWhere most of the course happens

Sources: OpenAI terms of use, Google Gemini help, GitHub terms of service and Anthropic consumer terms, as recorded in our AI tools age audit; read on 9 November 2026. Terms change; check the vendor's own page. Full list on our AI tools age guide.

There is a teaching reason as well as a legal one. A student who asks a chatbot to write a model never learns why the model fails. A student who writes the twenty lines below, and then breaks them on purpose, learns something no tool can hand them: a sense of when to be suspicious of a number.

A real lesson

How do you know a model is good? A lesson on test-set leakage

Here is a lesson from our teen course. Two models predict whether a student passes a weekly quiz. One of them looks better on a test set that quietly shares examples with its training data. On a clean test set, the verdict flips.

  1. The data. Twelve invented training rows: how many practice problems a student solved that week, on how many days they practised, and whether they passed. Two test sets: a "leaky" one with five of its six rows copied from training, and a "clean" one built from new rows.
  2. Model A, the memoriser. It stores every training example exactly. Asked about an example it has seen, it repeats the answer; asked about anything new, it says "fail".
  3. Model B, the rule. It tries every threshold from 0 to 10 on the number of problems solved and keeps the one that gets most training rows right. It learns: pass if 4 or more problems.
Two models, two test setsPython
# (problems solved this week, days practised) -> passed the quiz? Invented by us.
train = [((1, 1), 0), ((2, 1), 0), ((2, 3), 0), ((3, 2), 0), ((4, 4), 1), ((4, 1), 0),
         ((5, 3), 1), ((6, 2), 1), ((6, 5), 1), ((7, 1), 0), ((8, 4), 1), ((9, 5), 1)]
leaky_test = [((2, 3), 0), ((7, 1), 0), ((8, 4), 1), ((4, 1), 0), ((6, 2), 1), ((5, 4), 1)]
clean_test = [((5, 4), 1), ((3, 3), 1), ((7, 3), 1), ((1, 2), 0), ((6, 1), 1), ((3, 1), 0)]

memory = dict(train)          # remember every example exactly

def memoriser(x):
    return memory.get(x, 0)   # never seen it? say "fail"

def best_threshold(data):
    hits = {t: sum((p >= t) == bool(y) for (p, d), y in data) for t in range(11)}
    return max(hits, key=hits.get)

T = best_threshold(train)

def rule(x):
    return int(x[0] >= T)

def accuracy(model, data):
    return round(sum(model(x) == y for x, y in data) / len(data), 2)

print("threshold learned:", T)
for name, model in [("memoriser", memoriser), ("rule", rule)]:
    print(name, "| train", accuracy(model, train), "| leaky test",
          accuracy(model, leaky_test), "| clean test", accuracy(model, clean_test))
print("duplicates in leaky test:", sum(x in memory for x, y in leaky_test), "of", len(leaky_test))

Run with Python 3, it prints the four lines below. Every number in the data is ours, chosen to make the effect easy to see.

What it printsOutput
threshold learned: 4
memoriser | train 1.0 | leaky test 0.83 | clean test 0.33
rule | train 0.83 | leaky test 0.67 | clean test 0.83
duplicates in leaky test: 5 of 6

On the leaky test set the memoriser wins, 0.83 to 0.67. On the clean test set the rule wins, 0.83 to 0.33.

What went wrong

The memoriser scores 1.0 on its own training data, which proves nothing; any lookup table can do that. The leaky test set looks like a fair exam, but five of its six questions were already in the textbook, so the memoriser simply recalls them.

Two of those copied rows are odd cases, a student who solved 7 problems and still failed, another who solved 4 and failed. The rule gets those "wrong" because it learned the general pattern; the memoriser gets them "right" because it remembers the exceptions. A leaky test rewards remembering exceptions.

What the clean test shows

On six new students, the memoriser falls to 0.33: it has never seen any of them, so it answers "fail" every time. The rule, which learned a pattern instead of a list, keeps its 0.83.

Google's guidance says a good test set "contains no duplicates from the training set" and recommends deleting duplicates before testing. The last line of our program, duplicates in leaky test: 5 of 6, is the check that would have caught the problem before anyone believed the score.

  1. Predict first. Before running the code, write down which model you expect to win on each test set, and why.
  2. Break it differently. Change the memoriser so that unseen examples get "pass" instead of "fail". What happens to its clean score, and what does that tell you about guessing?
  3. Wear out the test. Try thresholds by hand and pick the one that scores highest on the clean test set. Explain in two English sentences why that score is now too optimistic.
  4. Write the report. One paragraph, in English, recommending a model to a teacher who will use it. Say what data you tested on and what you are still unsure about.

When it matters

Why does honest testing matter outside the classroom?

Because models are already used to make decisions about people, including students. Peru's 2025 AI regulation treats some of those uses as high risk.

Article 24.1 of DS 115-2025-PCM, the regulation of Peru's AI law published in El Peruano on 9 September 2025, lists uses of AI it considers high risk. One of them is evaluation in the processes of children and adolescents in education, including "evaluación de resultados de aprendizaje", the assessment of learning outcomes, unless the system plays a complementary role "y no reemplace la evaluación pedagógica humana", that is, unless it does not replace a teacher's judgement.

We quote this as context, not as legal advice. Its relevance to a teenager is simple: if a model ever marks your work, predicts your results or flags you during an exam, someone should have tested it on clean data and a human should still be checking. A student who has seen a memoriser win on a leaky test set understands why, and is better placed to ask the right questions.

Questions worth asking of any model

What data was it trained on, and who chose it? Was it tested on people like me? Were any of the test examples also in the training data? Who checks its decisions, and can I ask for a person to look again?

What we ask of our students

Every project ends with a short written note: the data, the test, the score on unseen examples, and one way the model could fail. It takes ten minutes and changes how students think about every number they see afterwards.

Competitions

Can AI help in programming olympiads in Peru?

Not during the contest. UTEC runs the OPI, Peru's computing olympiad, at no cost to any secundaria student in the country, and its 2025 rulebook forbids generative AI for solving the problems.

What the OPI 2025 rules say

Competitors must reason and code alone. The same rules explain that the top four of the second phase were selected for the International Olympiad in Informatics, the IOI, held in Sucre, Bolivia, from 27 July to 3 August 2025, and that Peru has taken part in the IOI since 2022.

Participants had to be at most 20 years old on 1 July 2025. Neither UTEC nor the OPI has any link with us, and we never reproduce its problems; we write our own practice problems and label them as ours.

How an AI course helps anyway

Olympiad problems reward clear thinking about data and edge cases, which is exactly what careful model testing trains. A student who habitually asks "what have I not tested?" writes better competition code as well.

Students who want competition depth can add our problem solving and DSA course, below, alongside the AI course. For a gentler start, our clases de programación para adolescentes page covers coding for teens in general.

In English

Is my English good enough for an AI course in English?

If you can follow a YouTube tutorial in English, yes. Our classes use clear, unhurried English, and the vocabulary of AI (data, model, train, test, accuracy) is the same short list of words used everywhere.

Why English helps

The documentation, error messages and research papers in AI are overwhelmingly in English. A student who learns "overfitting" and "test set" in English can read the original sources the day they need them.

EF's English Proficiency Index 2025 gives Peruvian adults aged 18 to 20 a score of 469, in the "low" band, while those aged 21 to 25 score 540, "moderate". Those are adults who chose to take EF's test, not school students, but the pattern suggests the late teens are a good time to build working English.

How we support it

Mentors repeat instructions in simpler words and write key terms in the chat. Students explain their code aloud in short sentences, and their written project notes are in English too, corrected gently.

We do not teach in Spanish. For a student who finds the first weeks hard, one-to-one classes let the mentor slow down without holding back a group. The free first class will show quickly whether it suits.

Progress

From first program to honest model: the stages

Our own stages, used to place a student and to report progress. They are not a school grade or an official level.

AI course stages for teenagers (ours)
StageThe student canEvidence we look for
1. Python basicsWrite loops, functions and listsA working program they can explain
2. DataLoad, clean and summarise a small datasetA chart with an honest caption
3. First modelTrain a threshold or tree by handThe model's rule, in their own words
4. Fair testSplit data and check for duplicatesA clean test score, reported alone
5. CompareChoose between two models fairlyA short report in English
6. JudgeExplain where a model could harm someoneA list of its likely failures

For parents

Stage 4 is the lesson above. Once a teenager reaches it, they tend to question scores, claims and statistics far beyond computing.

You will receive notes on what was built and understood, not just attendance.

For students

A high score on data the model has seen is not evidence. Ask what it was tested on.

When two models disagree, the interesting question is why.

More courses

Nine more courses for secundaria students

Each is taught live and in English, with one mentor across the months.

I

Programming foundations

From 13

AIT / BASE / 01

Python for teens

The whole language, start to advanced, before or beside AI.

Open the syllabus

AIT / BASE / 02

Java for teens

Objects and classes, for students heading to engineering.

Open the syllabus

AIT / BASE / 03

PCEP and PCAP preparation

The Python exams MINEDU's pilot also points to.

Open the syllabus
II

Going deeper

14 and up

AIT / DEEP / 01

Problem solving and DSA

Algorithms and data structures for olympiad-style thinking.

Open the syllabus

AIT / DEEP / 02

Orange data mining

Machine learning with visual blocks, then with Python.

Open the syllabus

AIT / DEEP / 03

Vibe coding for teens

Building with AI help while reading every line.

Open the syllabus
III

Projects and school exams

Ages 15 to 18

AIT / MORE / 01

Hackathon prep for teens

Twelve weeks of building before a friendly event.

Open the syllabus

AIT / MORE / 02

IB DP Computer Science

For Diploma students; the school marks all assessed work.

Open the syllabus

AIT / MORE / 03

AI agents with Copilot Studio

One to one only; a parent accepts Microsoft's terms.

Open the syllabus

How classes run

How do live AI classes for teens work?

Any day, any time, flexible: choose a batch, or for a very personalised plan choose 1-on-1 at the hour that suits your family.

Early morning

Before school in Peru is evening in India, a comfortable class hour.

After school

Later hours in Peru are possible too; we match a mentor to them.

Just you

One student and one mentor, at a pace set by the student.

Code every class

Students type and run their own programs live.

Test before trust

No score counts until it is on unseen data.

Clear English

Short sentences, key words written in the chat.

Teen groups only

Students learn with others of a similar age.

Projects with notes

Each project ends with a short written report.

Age-legal tools

AI tools only where the terms allow.

Student work

Projects by older students

Teenagers who stay with it build work like this; see more on our student labs page.

NutriLife AI nutrition coach project screenshot

AI and ML

NutriLife

An AI nutrition coach that reads what you eat and works you toward a target.

by Bhavya · Open it

Misti AI chatbot for maths and coding screenshot

AI and ML

Misti

A chatbot that answers mathematics and programming questions, built and deployed by a student.

by Harshit · Open it

GuardianX AI internet safety assistant screenshot

AI and ML

GuardianX

An assistant that helps a young person recognise unsafe situations online.

by Vivaan · Open it

SkyCast weather forecast application screenshot

Web app

SkyCast

A weather forecasting site with live conditions for any location.

by Krish

Fees

Fees for the teen AI course

Our fees are in US dollars, charged to a card. A Peruvian bank converts each payment to soles at its rate for that day, so expect small month-to-month differences on the statement.

Free first class

Free

no card required

  • A free first class
  • A short Python task and a first model
  • No card to book
Book it

Group batch

$100 USD

a month, eight live classes

  • Fifteen to twenty teenagers
  • The same mentor throughout
  • Two live classes each week
  • Month by month; cancel any time
Start here

One to one

$150 USD

a month, eight live classes

  • One student, one mentor
  • Projects built around the student's goals
  • Class times the family decides
Enquire

What families say

Rated 4.9 across 547 Google reviews

Reviews left on Google by students and parents, unedited.

★★★★★

"The one step solution for my son. Modern Age Coders make learning coding so simple that kids love it. The teachers explain complex concepts clearly with practical exercises and interactive content."

Ria Mukherjee

Parent

★★★★★

"Modern Age Coders has been a game-changer for me. I struggled to grasp IT concepts and coding before joining, but their classes transformed everything. I can now confidently write complex programs with ease."

Samriddha Mondal

Student

★★★★★

"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development."

Vansh Agarwal

Student

★★★★★

"My child Dhairya is really enjoying the Modern Age Coders classes. This is his first online class and he eagerly looks forward to it. I can already see his improvement, and the teachers are very cooperative."

Sonam Oswal

Parent of Dhairya

★★★★★

"Modern Age Coders have wonderful teachers who teach in a clear, easy and practical way. The teacher boosts students' confidence and inspires them to learn without hesitation."

Sonu Goyal

Parent

★★★★★

"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding."

Ritu Kedia

Parent

Students and parents ask

AI course for teens: common questions

Are the classes in Spanish?

No. The course is taught live online in English and only in English. Mentors keep their pace steady, rephrase when something is unclear and adjust to each student's accent, so a teenager with school-level English can usually follow and improves quickly.

Does my teenager need to know how to code already?

No. The core course starts with Python from the beginning. Students who already code skip ahead after the trial class. School maths, such as percentages and simple graphs, is all the maths needed to start.

Will my teenager use ChatGPT in the course?

Only with your permission and only where the terms allow. OpenAI's terms require a parent or guardian's permission under 18, and Anthropic's Claude is for adults only, so it is never used by students under 18. Most of the course uses Python itself, with no AI account.

What will my teenager actually build?

Small models trained on datasets they can understand, such as a threshold rule or a decision tree, then projects with real libraries. Every project is tested on examples the model has never seen and written up in a short English report.

At what times are classes held from Peru?

Any day, any time, flexible: choose a batch, or for a very personalised plan choose 1-on-1 at the hour that suits your family. Peru is on UTC minus 5 all year and India on UTC plus 5:30; 07:00 in Lima, before school, is 17:30 in India.

Does secundaria in Peru teach AI?

Not as a subject. The CNEB has no computing area in secundaria. MINEDU has run initiatives such as the Fábrica de Programadores Python pilot for 1,000 students in Lima Norte (announced July 2025) and an annual Hackathon, but most students meet neither.

Can the course help with the OPI programming olympiad?

It builds the right habits, but the OPI 2025 rules ban generative AI during the contest, so students must solve problems alone. For competition depth we suggest adding our problem solving and DSA course. We are not connected with UTEC or the OPI.

Is there a certificate?

Students who complete a course receive our own certificate of completion. It is not a Peruvian school qualification, and we are not connected with MINEDU, SUNEDU or any school. The projects and reports a student builds are usually the stronger evidence.

What equipment does a student need?

A laptop or desktop computer, a stable internet connection and a headset with a microphone. A phone is fine for watching but not for writing Python. All the software we use in the first months is free.

Can my teenager switch from a group to one-to-one?

Yes. Some students start in a group and move to one-to-one before exams or for a big project, and some do the reverse. Tell us and we arrange it from the next month.

Read next

More for secundaria students

Related pages on our site.

Coding and AI classes in Peru

Our full offer for Peru, by age and subject.

Programación para 5to de secundaria

A Spanish guide for the final school year.

AI tools age guide

Minimum ages for more than a hundred tools.

We teach you how to think

Reasoning before tools, in every course.

Student labs

Projects our students have built.

How we teach

What happens in a live class.

More pages for Peru

Part of the guide: Every Peru page

Subjects and courses

Subjects and courses (in Spanish)

More guides: Coding and AI classes in Peru · En español: clases de programación en inglés · Start here: English-medium coding in Lima · By age and stage: AI and coding for professionals · Exams, olympiads and school holidays: IB computer science tutor · Guides to choosing: An English-taught option to Crack The Code · Regions of Peru: Amazonas region · Cities: Arequipa: coding for every age · Lima and Callao districts: Programación para niños en Ancón

Start here

Try a free AI class

The trial is a real lesson: the student writes a few lines of Python, trains a tiny model and tests it. The mentor sees how they reason and how comfortable they are in English, then suggests where to start and whether a group or one-to-one fits better.

Curious about the format first? Read how we teach, or browse the syllabus of each course in our course list.

WhatsApp us · +91 91233 66161 · connect@modernagecoders.com

All teaching is live on video, in English, by mentors in India. Message us on WhatsApp in English for a quick answer.

No card, no commitment, no cost.

Want a class sooner? Book a priority demo

A priority demo is a full class of about 45 to 60 minutes. After the class the mentor sends a written skill report and a personal learning roadmap on WhatsApp. If you enrol, the demo fee is adjusted against your first month's fee. Share your demo Order ID with us on WhatsApp and we take it off.

Book a priority demo

Keep exploring Modern Age Coders

Related courses and guides

Learn more

Free resources

From the blog

Start here

Free classPriority demoCall us