CBSE Class 10 · Artificial Intelligence, code 417 · Paper analysis and answer key

Class 10 AI paper analysis: the 2026 paper taken apart, ready for 2027.

This class 10 AI paper analysis covers the CBSE 417 main paper of 27 February 2026: how hard each section was, the answer to every objective item put in our own words, and what the official marking scheme rewarded in each written answer. It stays at this address, and on 24 February 2027 the same page gets that day's paper. In our Monday and Wednesday batch, lessons like these shape every mock before the next paper.

Live online · 2026 paper analysed from CBSE's own marking scheme · answer key in our words · updated on 24 February 2027 · first class free

Start here

Courses behind this analysis

The 417 course whose Class 10 batch turns each paper into practice, the Python course for the practical half the paper never tests, and the visual machine learning course for students who want models to make sense rather than be memorised.

Ans. The short version

The CBSE Class 10 AI (417) paper of 27 February 2026 was moderate overall. Section A, 24 marks of objective items with no negative marking, was mostly direct but had a few traps in two-statement and assertion-reason items. Section B, 26 marks, rewarded exact terms: a confusion matrix case, Bag of Words on two short documents, and AI, ML and DL with a Venn diagram were the four-mark anchors, and teachers called the matrix and Bag of Words items scoring. The 2027 paper is on 24 February 2027, 10:30 AM. Modern Age Coders prepares students live, Monday and Wednesday at 9:00 PM IST, or one to one.

Q1. Was it easy?

Class 10 AI paper analysis 2026: was the AI paper easy?

It was moderate: easy to start, with marks lost in a handful of precise items rather than in hard topics. The paper kept the familiar frame, 21 questions with 15 to answer in two hours for 50 marks, and covered every theory unit. Our section-by-section reading of set 4, from CBSE's own marking scheme, is below.

SectionMarks counted (printed)Our difficulty ratingWhere the marks were won or lost
A, Q1: employability skills objective4 (6)EasyTwo items needed a written line, a definition and a fill-in, not just a choice
A, Q2 to Q5: AI objective20 (24)Easy to moderateOne two-statement item, two assertion-reason items and a domain question with an unexpected answer
B, Q6 to Q10: employability two-markers6 (10)EasyLists with partial marks, so a full list mattered more than neat prose
B, Q11 to Q16: AI two-markers8 (12)ModerateEach wanted a difference or a definition plus an example; half a mark rode on each example
B, Q17 to Q21: AI four-markers12 (20)ModerateTwo calculation-style questions were the safest picks; the theory four-markers needed structure

What teachers said on the day

Teachers quoted by India TV and Careers360 on 27 February 2026 described the paper as balanced and within the syllabus, with an overall moderate level. They called the four-mark Bag of Words and confusion matrix questions scoring, said most objective items were application-based, and noted that a couple of objective items had closely related options. One school also reported that two written questions needed careful interpretation.

How we read it

  • Choice was generous. 72 marks were printed for a 50-mark paper, so a student could skip the item they disliked in every block.
  • Precision decided the top band. Close options and exact terms separated a high score from a full one, not knowledge of extra topics.
  • No unit was skipped. Every theory unit appeared in both sections, so dropping one unit was a gamble.
  • Only set 4 was read. Other sets carried different items of the same type; our ratings describe set 4.

Q2. What were the right answers?

CBSE class 10 AI answer key 2026 for Section A, set 4, in our words

Every objective item of set 4 (question paper code 104/4), with what it tested described in our words and the answer CBSE's marking scheme keyed. We do not reprint the questions or options; the answer is given as the idea, so it can be checked against any copy of the paper.

Q1, employability skills (any 4 of 6)

ItemTestedKeyed answer
1 (i)Spotting the sentence type that gives an orderThe command sentence
1 (ii)Which SMART quality small steps give a big goalAchievable
1 (iii)A one-line meaning of time managementAny sound definition
1 (iv)What a strong password containsLetters, numbers and special characters together
1 (v)A myth about entrepreneurs, as a fill-inBorn, not made (other fair pairs accepted)
1 (vi)The odd one out among UN goalsPopulation

Q2, AI (any 5 of 6)

ItemTestedKeyed answer
2 (i)Naming a fruit in a photo with one labelClassification
2 (ii)A colour word used for angerMeaning depends on context
2 (iii)The purpose of ethical frameworksAvoiding unintended harm
2 (iv)Two statements: overfitting, and testing on training dataFirst true, second false; "both correct" also accepted
2 (v)Teaching a car to parkReinforcement learning
2 (vi)Project-cycle stage that tests on fresh dataEvaluation

Q3, AI (any 5 of 6)

ItemTestedKeyed answer
3 (i)The costliest error for a self-driving safety systemFalse negative
3 (ii)Pixel value range in a byte image0 to 255
3 (iii)"Bought this, also bought that" suggestionsAssociation model
3 (iv)What quietly shapes design decisionsIntuition and values
3 (v)Speech turned into text liveAuto-generated video captions
3 (vi)Why a testing set existsTo evaluate the model's accuracy

Q4 and Q5, AI (any 5 of 6 in each)

ItemTestedKeyed answer
4 (i)An alarm for a fire that is not thereFalse positive
4 (ii)Assertion and reason on bioethicsAssertion false, reason true: bioethics is sector-based
4 (iii)Which scenario is regressionForecasting a temperature
4 (iv)Pixels and resolutionResolution is the number of pixels
4 (v)Definition of precisionTrue positives over all predicted positives
4 (vi)Chatbot that needs coding and large databasesSmart bot
5 (i)Domain for a price comparison siteStatistical Data
5 (ii)Assertion and reason on lowercasing textBoth true, reason explains it
5 (iii)Computer vision against image processingComputer vision is the wider field
5 (iv)True or false on what error measuresTrue
5 (v)Reading satisfaction from reviewsSentiment analysis
5 (vi)Correct predictions from four matrix counts800 (200 + 600)

Two items are worth a second look. In 2 (iv), the scheme accepted a second answer because each statement could be judged true on its own wording, which shows CBSE adjusting the key when an item reads two ways. In 4 (ii), the trap was the word "value-based": bioethics belongs to one sector, healthcare, so the assertion fails while the reason holds. Original practice in these formats is on assertion reason questions for Class 10 AI and the Class 10 AI MCQ bank.

Q3. How were the written answers marked?

AI question paper solution: what CBSE rewarded in each written answer

The marking scheme gives examiners value points, not model essays, and it splits the marks inside each answer. Knowing that split is the real solution to Section B, because it tells you what to write first. Here is set 4, described in our words.

QuestionWhat it asked, in our wordsHow the marks were splitOur note
6The seven Cs of communication2 for all seven, 1.5 for five or six, 1 for three or four, 0.5 for one or twoA pure list; every C named was worth something
7Emotional intelligence: meaning, and two of its skills1 for the meaning, 0.5 per skillEasy if the skill names were learnt
8 to 10Uses of ICT skills; qualities of an entrepreneur; sustainable development defined1 per relevant point, or 2 for a sound definitionGenerous; most students lose nothing here
11Sector-based against value-based frameworks, an example of each1 for a difference, 0.5 per exampleBioethics as the sector example; rights, virtue or utility as the value one
12Two features of a classification model1 per featureSupervised, labelled, discrete classes
13How a computer stores a colour image2 for a correct accountThree channels, each pixel holding three values from 0 to 255
14Two differences between supervised and unsupervised learning1 per difference; 0.5 per example if only examples givenLabels, and what the model is trying to find
15The train-test split2 for a correct accountMention the unseen data and a ratio such as 80 to 20
16Stemming against lemmatisation, shown on an irregular plural1 for the difference, 0.5 for the stem, 0.5 for the lemmaThe stem may not be a real word; the lemma always is
17AI, ML and DL told apart, with a labelled Venn diagram1 per term, 1 for the diagramNested circles, DL inside ML inside AI
18Two scenarios, each matched to an AI domain with a reason1 for the domain, 1 for the reason, in eachEssay feedback was NLP; a vehicle-counting camera was computer vision
19A cyber-attack detector: matrix, true negatives, precision2 for the matrix, 1 for TN, 1 for precisionEither matrix orientation accepted if the labels were right
20CNN and ANN expanded and defined; two boxes in a network diagram0.5 per full form, 1 per definition, 0.5 per boxThe boxes were the input and hidden layers
21Bag of Words on two short documents, all four steps1 per stepSteps three and four could be combined without loss
Question 19, worked from the scenarioattack = positive
1500 network activities tested.
Correct "attack": 1000      TP = 1000
Correct "no attack": 250    TN = 250
Said attack, was not: 200   FP = 200
Said safe, was attack: 50   FN = 50

               Reality: Yes  Reality: No
Predicted Yes    1000          200
Predicted No       50          250

Precision = 1000 / (1000 + 200)
          = 1000 / 1200 = 0.833

Not asked, but worth knowing: recall 1000 / 1050 = 0.952 and accuracy 1250 / 1500 = 0.833. Here precision and accuracy round to the same value, which is a coincidence, not a rule.

Question 21's method, on our own sentencesfour steps
Doc 1: Robots sort mail quickly.
Doc 2: Mail reaches people quickly.

1. Normalise: lowercase, drop the full stop
   [robots, sort, mail, quickly]
   [mail, reaches, people, quickly]
2. Dictionary: robots, sort, mail,
   quickly, reaches, people
3. Vector for Doc 1: 1 1 1 1 0 0
4. Table:     ro so ma qu re pe
   Doc 1      1  1  1  1  0  0
   Doc 2      0  0  1  1  1  1

The 2026 paper's pair shared three of their words, so its dictionary had five entries and each document vector had four ones and one zero. One mark per step means a student who stops after the dictionary still keeps two.

The scheme also gave half a mark for writing the precision formula alone and told examiners not to penalise the arithmetic, so the formula should always go down first. How these value points are applied across every question type is on the CBSE Class 10 AI marking scheme page; more matrices with fresh numbers are on confusion matrix numericals for Class 10 AI.

Q4. What does it mean for 2027?

AI exam review: six lessons from the 2026 paper for the 2027 candidate

A paper analysis is only useful if it changes how the next student prepares. These are the habits the 2026 paper rewarded, each one now built into our mock reviews.

Learn the frameworks by type

Two items turned on whether a framework is sector-based or value-based. Students who knew bioethics only as "an ethics example" fell into the assertion trap.

Expect the unusual domain

A price comparison website belongs to Statistical Data, not NLP. Domain questions reward reasoning from the data the system uses, not from what it looks like.

Pick the calculation four-markers

The matrix and Bag of Words questions gave a mark per step, so a methodical student was close to certain of eight marks from two choices.

Draw what can be drawn

A labelled Venn diagram carried a full mark, and two network boxes another. A clean sketch is the quickest mark in Section B.

Give the example every time

Several two-markers split their marks between a point and an example. An answer with the right idea and no example stopped at half.

Read both statements on their own

Two-statement items ask about each statement separately. Judge statement one, then statement two, then choose; do not let one colour the other.

Every theory unit, with written answers of each length, is practised on CBSE Class 10 AI important questions, and full timed papers on the sample paper and mock tests page.

Q5. What happens on 24 February 2027?

Class 10 AI paper analysis 2027: what this page adds on exam day

The 2027 AI paper is on Wednesday 24 February 2027, from 10:30 AM to 12:30 PM IST, on the same morning as Computer Applications and Information Technology. This address does not change: the 2027 analysis is added here, in stages, and the 2026 analysis stays below it for comparison.

WhenWhat is addedSource
Exam day, afternoonThe section-by-section shape of the paper and our first difficulty notesStudents and teachers we speak to after the paper
Within the following daysAn answer key for the objective items, written in our words, and solution notes for the written questionsOur teachers working the paper in full
When CBSE posts its marking schemeAny item where the official key differs from ours, corrected and notedCBSE's marking scheme page
If CBSE issues a 2026-27 sample paperHow the real paper compared with itcbseacademic.nic.in; the latest 417 sample paper there is still the 2025-26 one

What will not be added

No copy of the question paper and no reproduced question text. The analysis and the answer key describe each item in our words, as on this page. We also do not post "expected" questions before the paper: the 2027 paper is unpublished until the morning it is sat.

Until then, in the live batch

Monday classes take each theory unit through the shapes analysed above, and Wednesday classes keep the practical file and project moving. From January, full papers are written at 10:30 AM on weekends to match the exam slot, then reviewed live against the value points. The run-in plan is on the CBSE Class 10 AI revision plan.

Batch or personalised: you choose

Practise for the next paper in the Monday and Wednesday batch, or one to one

Both formats work from the same papers, the same marking and the same teachers. The batch gives the whole syllabus in order with classmates writing the same mocks; one to one goes straight to the items your child keeps missing. Monthly billing, no admission fee, stop at the end of any month.

Option 1 · Fixed batch timings

Group batch, Monday and Wednesday 9 PM

₹1,499

per month, two live classes a week

  • Monday and Wednesday, 9:00 PM IST, live on video
  • Fifteen to twenty Class 10 AI students, one teacher
  • Objective items timed, each option explained
  • Written answers marked against value points on screen
  • Morning mock papers from January
Book a free demo in this batch

Option 2 · Choose your own timing

Personalised one to one

₹4,999

per month, one private class a week, at a time set with you

  • Days and time picked by you, in your own time zone
  • Practice drawn from the item types your child misses
  • Four-mark answers rewritten until every point lands
  • Mocks scheduled around school pre-boards
  • Same CBSE 417 syllabus and teachers
Ask for a personalised plan

A student who already scores well but drops marks on close options and missing examples often closes the gap in a few one to one sessions. A student still meeting some units for the first time does better in the batch, where the full syllabus is taught before the papers begin. Fees are shown in your own currency.

Studying outside India?

The same AI paper for CBSE students in the Gulf and abroad

CBSE schools in the UAE, Saudi Arabia, Qatar or Kenya sit the same 417 paper on the same morning, so this analysis applies to them unchanged. Here is the batch hour in local time.

Where you liveMon and Wed 9:00 PM ISTBest format
India (IST)
every Indian city
9:00 PMBatch fits
UAE and Oman
Dubai, Abu Dhabi, Sharjah, Muscat
7:30 PMBatch fits
Saudi Arabia, Qatar, Kuwait, Bahrain
Riyadh, Jeddah, Dammam, Doha, Kuwait City, Manama
6:30 PMBatch fits
Nepal
Kathmandu
9:15 PMBatch fits
Ethiopia and East Africa
Addis Ababa and East Africa Time
6:30 PMBatch fits
Singapore and Malaysia
Singapore, Kuala Lumpur
11:30 PMChoose 1-on-1

If 9:00 PM IST is late where you are, one to one runs at a local hour of your choosing. Families abroad will find more on CBSE Class 10 AI online classes in the Gulf and abroad.

What families say

Rated 4.9 across 547 Google reviews

Real reviews from real families. We neither write nor commission them.

★★★★★

"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

The rest of the Class 10 series

More pages for Class 10 board students

Every page in the CBSE Class 10 AI (417) series, grouped by what you need: classes, each unit, question practice, notes and revision, and the exam itself.

ICSE Class 10 Computer Applications (Java): The Tue and Wed ICSE Java batch ICSE Computer Applications syllabus ICSE Java programs practice ICSE CA important questions

Questions about the AI paper

What students ask after the class 10 AI paper

Was the CBSE class 10 AI paper 2026 easy?

It was moderate. Teachers quoted on the day called it balanced and within the syllabus, and the objective section was mostly direct. Marks were lost on precision rather than hard topics: a two-statement item, two assertion-reason items, an unexpected domain question, and written answers that left out an example or a step.

Where can I find the CBSE class 10 AI answer key 2026?

This page gives the answer to every objective item of set 4 (question paper code 104/4), with what each item tested described in our own words. It is based on CBSE's own marking scheme for the 2026 main examination, which CBSE has posted on its marking scheme page. Other sets carried different items of the same types.

When is the class 10 AI paper analysis 2027 coming?

On the day of the paper. CBSE's 2027 date sheet puts the Artificial Intelligence (417) paper on Wednesday 24 February 2027, from 10:30 AM to 12:30 PM IST. This page will add the section-by-section shape and difficulty notes that afternoon, then an answer key in our own words in the following days.

How many questions were in the 2026 AI paper?

Twenty-one, of which students answered fifteen, in two hours for 50 marks. Section A had five objective questions of six parts each, answer any four or five parts. Section B had sixteen written questions, of which ten were answered: three employability two-markers, four AI two-markers and three AI four-markers.

Which questions were the most scoring in the AI paper?

Teachers named the four-mark Bag of Words and confusion matrix questions, and the marking scheme supports that: Bag of Words gave a mark per step, and the matrix question split its four marks between the matrix, one cell and precision. A methodical student could expect most of those eight marks.

Did the 2026 AI paper have assertion reason questions?

Yes. Set 4 had two assertion-reason items in Section A, one on bioethics and one on lowercasing text, plus a two-statement item on overfitting. CBSE has not said how many such items a 2027 paper will carry, so treat them as one of several Section A formats.

Was there any mistake or change in the AI answer key?

In set 4, the two-statement item on overfitting was keyed as first statement true and second false, but the marking scheme also accepted "both correct", noting that each statement could be read as true on its own. No other multiple-choice item in set 4 was keyed with a second accepted option.

Did the AI paper ask Python questions?

No. Statistical Data and Advance Python are assessed only in the practical, so the 2026 theory paper had no code. Statistical Data appeared only as an AI domain, in a question about which domain suits a price comparison website. Python is examined through the practical file, practical exam and viva.

Is this page a copy of the question paper?

No. We do not reproduce the paper, its options or its wording. Each item is described in our own words, the answers are stated as ideas, and worked methods use our own example sentences except where a figure, such as the confusion matrix counts, is needed to show the arithmetic.

How do your classes use paper analysis like this?

Each item type from the real papers becomes a practice routine: timed objective sets with every option explained, written answers marked against the value points on screen, and full morning mocks from January. Classes run live on Monday and Wednesday at 9:00 PM IST, or one to one, and the first class is a free demo.

Start

Try a paper-style class free

The free first class works through one block of objective items and one four-mark question in the shapes analysed above, with the teacher marking against the CBSE value points on screen. Afterwards the mentor tells you plainly where your child would have lost marks on the 2026 paper and whether the batch or one to one is the better way to the next one.

Want every past paper in one place? See Class 10 AI previous year question papers.

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

Every class is live video. The form requests a callback and reserves nothing.

We call within a few hours. No card, no enrolment fee.

Want it sooner? Book a priority demo

A full live class of about 45 to 60 minutes at a slot you pick, with a written skill report and learning roadmap on WhatsApp afterwards. The fee is adjusted against your first month if you enrol.

Book a priority demo

Keep exploring Modern Age Coders

By class and age

Learn more

Free resources

From the blog

Start here