AICE / EXM / 01
PCEP and PCAP
The Python Institute's two entry exams, objective by objective.
Open the syllabusUnited States · Grades 9 to 12 · AI and Python exams
A tenth grader who has trained a first model often asks the same thing next: is there a certificate that says I can do this? There is more than one, and they are not alike. One comes from a testing company that sells to schools and test centers, some from a cloud provider for people starting careers, and one by a Python foundation that tests code rather than ideas. This page lays the four realistic options side by side using only what each issuer publishes: who may sit the exam, how long it runs, what it tests, how long the credential lasts, and what a teenager needs a parent to sign.
Live teaching since 2020 · 10,000+ students · We teach the skills; the issuers run the exams
The short answer
A US high school student can sit four realistic AI-related exams: Certiport's IT Specialist Artificial Intelligence (50 minutes), Microsoft's AI-901 Azure AI Fundamentals (pass mark 700), AWS Certified AI Practitioner (65 questions, ages 13 to 17 with a parent's consent) and the Python Institute's PCEP. Modern Age Coders teaches the skills behind them live online, ages 13 to 18, in a batch or one to one.
Learn the skills behind the badge
An exam measures a slice of what a student knows. These courses build the whole of it, and each one is taught live by a regular teacher.

AICE / 01
From about 13 upward. The Python Institute syllabus, taught line by line.
Open the syllabus →
AICE / 02
Ages 13 to 18. Models trained, scored and explained, the core of every AI exam.
Open the syllabus →
AICE / 03
Ages 14 to 18. Cleaning data and reading results, which the newer exam objectives stress.
Open the syllabus →The four exams
Every cell below comes from the issuer's own page, read on 17 October 2026. Where an issuer publishes nothing on a point, the table says so rather than guessing. Exam fees are listed on each issuer's page and change by country, so check them there.
| Exam | Issuer | What it tests | Length | Pass mark | Lasts | Under-18 rule |
|---|---|---|---|---|---|---|
| IT Specialist: Artificial Intelligence | Certiport, a Pearson VUE business | Using AI to solve problems: data collection, feature engineering, NLP, feasibility, deployment planning | 50 minutes, 45 to 50 questions | Set by Certiport | One-time issuance, valid from the date passed | Certiport says the series suits ages 14 and up |
| AI-901: Azure AI Fundamentals | Microsoft | AI concepts (40 to 45%) and building with Microsoft Foundry (55 to 60%) | Not stated on the exam page | 700 on a 1 to 1,000 scale | Fundamentals certifications do not expire | No age rule on the pages we read |
| AWS Certified AI Practitioner | Amazon Web Services | AI, ML and generative AI concepts and AWS AI services | 90 minutes, 65 questions | Set by AWS | This certification is valid for 3 years | 13 or older; 13 to 17 with a parent or legal guardian's consent |
| PCEP: Certified Entry-Level Python Programmer | Python Institute (OpenEDG) | Python itself: data types, loops, functions, exceptions | 40 minutes plus 5 for the agreement, 30 questions | 70% | 5 years | Not under 13; 13 to 17 with parental or legal guardian consent |
Three of the four test ideas about AI. Only PCEP tests whether you can write and read code, and it tests Python rather than AI. PCEP makes a natural first exam because it rewards exactly the work a beginner is already doing.
The Microsoft pass mark is a scaled score. Microsoft says a score of 700 "may not equal 70% of the points", because harder question sets need fewer points to pass. Do not convert it to a percentage in your head.
Microsoft strongly recommends registering with a personal account, warning that with a work or school account "your exam records will be lost if you leave your organization". A student who registers through a school email can lose the record at graduation.
For AWS, "Candidates must be at least 13 years old at the time of scheduling an exam", and the Python Institute says children under 13 are not allowed to test or receive certification even with parental consent. A parent signs for both until the student turns 18.
Sources, read 17 October 2026: Certiport IT Specialist and exam lengths; Microsoft Exam AI-901; AWS Certified AI Practitioner and the AWS Certification FAQs; Python Institute PCEP and its testing policies. We are not connected with Certiport, Pearson VUE, Microsoft, Amazon Web Services or the Python Institute.
Exam by exam
The comparison table flattens real differences. Here is the detail behind each row, in the issuers' words where the words matter.
Certiport describes candidates as "AI enabled learners" who are prepared for the professional use of AI. It expects approximately 150 hours of instruction and hands-on experience with the topic, and Certiport says the American Council on Education recommends three college credits per IT Specialist exam passed. A recommendation is not a guarantee: each college decides for itself.
Microsoft updated the English version of this exam on April 15, 2026. It asks for knowledge of Python coding syntax and familiarity with Azure resources, which is more hands-on than the word Fundamentals suggests. The weight sits on building: "Implement AI solutions by using Microsoft Foundry" makes up 55 to 60% of the exam.
AWS files this exam under its Foundational category and describes it as a starting point for people exploring AI and machine learning on AWS. It is the longest of the four at 90 minutes. Its credential lasts three years, the shortest stated validity of the four, and AWS lets a later exam renew it.
| Change | What Certiport says |
|---|---|
| Data objective split in two | Objective 2 became "Data Collection & Transformation" and "Feature Engineering" |
| Language models named | "Objective 4.1 now covers Natural Language Processing (NLP)", with algorithm families in 4.1.2 to 4.1.6 |
| New planning skills | Feasibility analysis, choosing visualization types for a data story, and planning for continuous integration and deployment |
| When | English, Spanish and French versions went live in June 2026 |
PCEP is also changing: the Python Institute lists PCEP-30-02 as the active version and PCEP-30-03 as scheduled for release on October 31, 2026 (as read on 17 October 2026), and says the update does not affect the five-year validity of credentials already earned. A student booking an exam this fall should check which version the voucher covers.
Sources: IT Specialist overview; the Certiport blog, 29 April 2026; Microsoft exam scoring and renewal pages; PCEP exam information.
Honest about value
Parents are right to ask what a certification is worth. The answer is narrower than the marketing around exams suggests, and more useful once it is clear.
That on one day, under supervised conditions, the student answered enough questions correctly on a published list of objectives. That is real information. The objectives are public, so anyone can see exactly what was measured.
For a younger teenager it is also a finish line. Preparing for a fixed date teaches revision habits that a self-paced tutorial never does.
It does not show that the student can build anything. An exam samples knowledge in a fixed time; a model on GitHub with an honest write-up of where it fails says more about ability than any badge.
It is not an admission ticket or a job offer, and no issuer claims it is. We make no such promise either: we teach the material, the student sits the exam, and the result belongs to them.
| Kind | Example | Who checks the work | What it tells a reader |
|---|---|---|---|
| Certification exam | AWS AI Practitioner, PCEP | The issuer, in a proctored test | The objectives were met on test day |
| Course certificate | The free CS50 Certificate for CS50 AI | Course staff, from submitted projects | The projects were completed to a set standard |
| Portfolio | A repository with code, data and a write-up | Anyone who opens it | What the student can actually do |
Harvard's CS50 AI offers its free certificate to anyone who submits and scores at least 70% on each of the course's projects; submitting needs an edX account, and edX says users must be at least 13. Federal policy has also moved toward credentials for teenagers: the April 23, 2025 executive order on AI education asks agencies to encourage dual enrollment so students can earn "industry-recognized AI credentials concurrent with high school education". Neither turns a certificate into a guarantee. Our own view of certificates is in a certificate earned, not given.
Choosing
There is no single right order, but there is a sensible default. This table is our advice, not any issuer's.
| Student | A sensible first exam | Why |
|---|---|---|
| Grade 9, some Python | PCEP | Tests code they already write; 30 questions in 40 minutes |
| Grade 10, school AI unit done | IT Specialist: AI | Certiport suits it to ages 14 and up; covers data, features and NLP |
| Grade 11, wants cloud skills | AI-901 or AWS AI Practitioner | Both expect comfort with a cloud console |
| Grade 12, building a portfolio | Whichever fits the project | The project matters more; the exam confirms the vocabulary |
| Any grade, no code yet | None yet | Learn first; an exam taken too early mostly measures guessing |
A worked example
Judging a model by more than one number is a core idea in AI fundamentals. Here is the idea worked through in code, on a dozen made-up emails. Both scripts are ours; we ran them with Python 3.13 and scikit-learn 1.7.2 on 17 October 2026.
# Our spam filter's report card (ours)
# 1 = spam, 0 = a normal email
actual = [1, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0]
predicted = [1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0]
pairs = list(zip(actual, predicted))
tp = pairs.count((1, 1)) # spam, caught
fp = pairs.count((0, 1)) # normal, flagged
fn = pairs.count((1, 0)) # spam, missed
tn = pairs.count((0, 0)) # normal, let through
print("TP", tp, "FP", fp, "FN", fn, "TN", tn)
print("accuracy ", round((tp + tn) / len(pairs), 3))
print("precision", round(tp / (tp + fp), 3))
print("recall ", round(tp / (tp + fn), 3))Output on our run: TP 4 FP 2 FN 1 TN 5, accuracy 0.75, precision 0.667, recall 0.8. The filter caught four of the five spam emails (recall 4/5) but two of the six emails it flagged were normal (precision 4/6).
# The same check with scikit-learn (ours)
from sklearn.metrics import (confusion_matrix,
precision_score, recall_score)
actual = [1, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0]
predicted = [1, 0, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0]
print(confusion_matrix(actual, predicted))
print(round(precision_score(actual, predicted), 3))
print(recall_score(actual, predicted))Output on our run: [[5 2] [1 4]], then 0.667 and 0.8. In scikit-learn's confusion matrix each row is the true class and each column the predicted class, so the top row reads 5 normal emails let through and 2 wrongly flagged.
Precision asks: when the filter shouts spam, how often is it right? Recall asks: of all the real spam, how much did it catch? scikit-learn defines them as tp / (tp + fp) and tp / (tp + fn), the same sums Script 1 does by hand.
Accuracy hides the trade. A filter that never flags anything would score 7 out of 12 here, about 58%, while catching no spam at all. That is our arithmetic, and it is the kind of trap a well-written question sets.
A student who has typed pairs.count((0, 1)) knows what a false positive is. One who has only read the definition usually mixes up the two error types under exam pressure.
Change one value in the predicted list and run it again. Watching precision move while recall stays put is worth a page of notes.
Practice set
These are ours, written in the spirit of the published objectives. They are not real exam items and no issuer has reviewed them. Answers are in the second column, so cover it first.
| Question | Answer and reason |
|---|---|
| 1. A model flags 10 photos as cats; 8 are cats. What is its precision? | 0.8: 8 correct out of 10 flagged. |
| 2. There were 16 cat photos and the model found 8. What is its recall? | 0.5: it found half of the real cats. |
| 3. A dataset has 95 normal transactions and 5 unusual ones. Why is 95% accuracy unimpressive? | Guessing normal every time scores 95% and finds nothing unusual. |
| 4. You turn a column of dates into day of week. Which objective is this? | Feature engineering: making a new input the model can use. |
| 5. Training accuracy is 99% and test accuracy is 70%. What is the likely problem? | Overfitting: the model memorized the training set. |
| 6. A chatbot gives a confident wrong date. What should a responsible user do? | Check a primary source before repeating it; fluent text is not proof. |
Questions 1 and 2 use the same model and land on different numbers, which is the whole point of the code above. Students who get question 3 right have understood why exams ask for more than accuracy. For the reasons models fail in the first place, see why machine learning models make mistakes.
Study plan
Our plan for a high schooler preparing for the IT Specialist AI exam or AI-901 alongside school. It assumes some Python already; without it, add the PCEP material first.
| Week | Focus | Proof of progress |
|---|---|---|
| 1 | Read the objective list; sort into known and unknown | A ticked list |
| 2 | Data: collecting, cleaning, splitting | A cleaned table and a train and test split |
| 3 | Features: what a model can learn from | Two new features and a sentence on each |
| 4 | Models: classification and regression | One trained model, scored |
| 5 | Evaluation: precision, recall, overfitting | The practice set above, all six right |
| 6 | Language and image AI, and their limits | One page in their own words |
| 7 | Responsible AI: bias, privacy, checking outputs | A short risk note on their own model |
| 8 | Official practice test, timed, then review | A score and a fix list |
Certiport expects about 150 hours of instruction and practice. Spread over eight weeks that is 150 / 8 = 18.75 hours a week, our arithmetic, which is too much beside school. So start earlier or take the exam later.
Our classes cover the skills in weeks 2 to 7 as part of a normal course. The official practice test and booking stay with the student and the issuer.
Read the score report by objective, not the total. One weak objective is a week of work; three weak ones mean the exam came too soon.
From learner to certified
Where a high schooler might start, and what moves them up a stage.
| Stage | In practice |
|---|---|
| 1. User | Uses AI apps and can describe what they do |
| 2. Coder | Writes Python and has trained one model on real data |
| 3. Evaluator | Can say why a model is wrong, with numbers |
| 4. Certified builder | Has passed an exam and shipped a project that matches it |
Stage 2 is where an exam starts to make sense. Before it, money spent on a voucher mostly buys a lesson in guessing.
Ask your teenager to explain their model to you. If they can, stage 3 is close.
Keep the exam objectives open beside your project. Each one you can point to in your own code is one you will not need to cram.
Plan for a pass, not a perfect score. The credential does not print the mark.
The course catalog
Each is live and online, with a regular teacher, taught in a batch or one to one.
AICE / EXM / 01
The Python Institute's two entry exams, objective by objective.
Open the syllabusAICE / EXM / 02
For students who already hold PCEP or its equivalent.
Open the syllabusAICE / EXM / 03
Using AI well, with the checking habits exams reward.
Open the syllabusAICE / SKL / 01
From a first script to programs worth testing.
Open the syllabusAICE / SKL / 02
Training and scoring real models.
Open the syllabusAICE / SKL / 03
pandas, charts and honest conclusions.
Open the syllabusAICE / BYD / 01
Machine learning on a visual canvas, then the step to Python.
Open the syllabusAICE / BYD / 02
Algorithms for the unseen question.
Open the syllabusAICE / BYD / 03
Where the data in every AI project lives.
Open the syllabusHow lessons run
Weekdays or weekends, flexible: choose your batch, or for a very personalised plan choose 1-on-1 at your convenience, weekend mornings or weekdays.
Weekdays
After school, in whichever US time zone you live.
Weekends
Weekend mornings suit students with busy weeks.
1-on-1
Paced to a booked exam date.
The same teacher from the first class to the exam.
We map each lesson to the list the student will be tested on.
Every idea is typed and run in class.
Our own practice items, never leaked exam content.
Live on video from home.
Fifteen to twenty in a batch, or one to one.
Student work
The kind of evidence that sits well beside a certificate. More on the student labs page.

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

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

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

Web app
A weather forecasting site with live conditions for any location.
Fees
Exam vouchers are bought from the issuer and are not part of our fees. What we charge is for teaching, a month at a time in US dollars, with no term contract; stop after any month.
Free first class
USD 0
no card required
Group batch
USD 100
a month, billed in US dollars
One to one
USD 150
a month, billed in US dollars
What families say
Reviews parents left on Google, copied without edits.
★★★★★
"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
Questions from families
Yes. Certiport's IT Specialist Artificial Intelligence exam is one, and Certiport says its IT Specialist exams suit ages 14 and up. AWS Certified AI Practitioner is open from 13, with a parent's consent until 18. Microsoft publishes no age rule for AI-901 on the pages we read, so check with Microsoft.
Yes, with consent. AWS says candidates must be at least 13 when they schedule an exam, and candidates aged 13 to 17 may test with the consent of a parent or legal guardian.
Certiport lists each IT Specialist exam at 50 minutes, with 45 to 50 questions for Artificial Intelligence.
Microsoft lists a passing score of 700 on a scale of 1 to 1,000. It is a scaled score, so Microsoft says it may not equal 70% of the points.
It depends on the issuer. Microsoft says Fundamentals certifications do not expire, AWS says its AI Practitioner certification is valid for 3 years, the Python Institute lists PCEP as valid for 5 years, and Certiport describes its exams as a one-time issuance.
No. PCEP tests entry-level Python: data types, loops, functions and exceptions. It pairs well with AI study because courses such as CS50 AI and libraries such as scikit-learn use Python.
We cannot promise that, and no issuer claims it. A certification shows that published objectives were met on one test day. Colleges decide for themselves what they value.
Microsoft strongly recommends a personal account, warning that exam records tied to a school or work account will be lost if the student leaves that organization. Use a personal email for any exam.
No. We teach the skills and the material in live classes. Students book and sit the exam directly with the issuer. We are not connected with Certiport, Microsoft, AWS or the Python Institute.
Weekdays or weekends, flexible: choose your batch, or for a very personalised plan choose 1-on-1 at your convenience, weekend mornings or weekdays.
The first class is free. After that families pay monthly for teaching, and the fees section of this page shows the figure in US dollars. Exam vouchers are separate and come from the issuer.
Elsewhere on this site
More for US high schoolers working on AI.
Every page we have written for US families.
Our own certificate track and its portfolio.
Which Python exam to sit first.
The evidence that sits beside an exam.
Which tools a teenager may use, from their terms.
Our live AI track for ages 13 to 18.
AI summer programs for US high schoolers in 2027, and an eight-week AI summer you can run at home.
AI literacy for K-5: what it means, what standards expect, and two activities.
ML and AI competitions for US high schoolers, 2026-27, from the official rules.
Start here
Tell us which exam your teenager is eyeing and what they have built so far. The teacher reads the objective list with them in the first class and says honestly how far away the exam is.
Curious how a class works first? Read how we teach, or open the course list, where each syllabus is public.
WhatsApp us · +91 91233 66161 · connect@modernagecoders.com
We teach from India over live video, so classes reach every US state without a commute. WhatsApp is the fastest way to reach us from the US.
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