Artificial Intelligence & Robotics MCQs for UPSC Prelims
44 practice questions on Artificial Intelligence & Robotics from the Science & Technology section of the UPSC Prelims syllabus.
44 come with a written explanation and 3 are actual previous year questions.
Try the sample set below - the answer stays hidden until you ask for it.
7 Easy23 Medium14 Hard3 from past papers
Sample questions
Q1
Previous year questionmedium
Consider the following statements :
I. It is expected that Majorana 1 chip will enable quantum computing.
II. Majorana 1 chip has been introduced by Amazon Web Services (AWS).
III. Deep learning is a subset of machine learning.
Which of the statements given above are correct?
AI and II only
BII and III only
CI and III only
DI, II and III
Show answer and explanation
Correct answer: C - I and III only
The Majorana 1 chip is designed to advance quantum computing using topological qubits, so statement I is correct, but it was introduced by Microsoft, not Amazon Web Services, so statement II is wrong. Deep learning, which uses multilayered neural networks, is indeed a subset of machine learning, validating statement III. Hence only I and III are correct.
Q2
Previous year questionhard
With the present state of development, Artificial Intelligence can effectively do which of the following?
1. Bring down electricity consumption in industrial units
2. Create meaningful short stories and songs
3. Disease diagnosis
4. Text-to-Speech Conversion
5. Wireless transmission of electrical energy
Select the correct answer using the code given below:
A1, 2, 3 and 5 only
B1, 3 and 4 only
C2, 4 and 5 only
D1, 2, 3, 4 and 5
Show answer and explanation
Correct answer: B - 1, 3 and 4 only
As per the official UPSC key the answer is (b) 1, 3 and 4 only: AI optimises industrial electricity use, performs disease diagnosis and powers text-to-speech systems. Statement 5 is wrong because AI does not achieve wireless transmission of electrical energy, which is a separate power-engineering problem. The UPSC key treated statement 2 (creating meaningful stories and songs) as not effectively achievable at the time, which is why option (d) is excluded. This question is disputed: several coaching keys mark (d) all five, but the official answer is retained here as (b).
Q3
Previous year questionmedium
Consider the following activities:
1. Spraying pesticides on a crop field
2. Inspecting the craters of active volcanoes
3. Collecting breath samples from spouting whales for DNA analysis
At the present level of technology, which of the above activities can be successfully carried out by using drones?
A1 and 2 only
B2 and 3 only
C1 and 3 only
D1, 2 and 3
Show answer and explanation
Correct answer: D - 1, 2 and 3
All three activities are feasible with current drone technology: agricultural drones spray pesticides over crop fields, drones safely inspect the craters of active volcanoes where humans cannot go, and specially fitted drones collect the blow (breath) spray from whales for DNA and microbiome analysis. Since each application has been demonstrated, the correct answer is 1, 2 and 3.
Q4
medium
In machine learning, 'supervised learning' is best described as a method in which:
AA model is trained on labelled input-output pairs to learn a mapping from inputs to outputs
BA model discovers hidden patterns in unlabelled data without predefined outputs
CAn agent learns optimal actions by receiving rewards and penalties from its environment
DA model is hard-coded with explicit rules by a human expert
Show answer and explanation
Correct answer: A - A model is trained on labelled input-output pairs to learn a mapping from inputs to outputs
Supervised learning trains a model on a dataset of labelled examples, where each input is paired with the correct output, so the model learns a mapping that can predict outputs for new inputs (e.g., classification or regression). Discovering patterns in unlabelled data is unsupervised learning; learning through rewards and penalties is reinforcement learning; and hard-coded human rules describe a rule-based expert system, not machine learning.
Q5
medium
In machine learning, 'overfitting' of a model refers to a situation where the model:
ALearns the training data including its noise too well and performs poorly on new, unseen data
BIs too simple to capture the underlying pattern in the data
CRequires excessive memory to store its parameters
DCannot be trained because of insufficient computing power
Show answer and explanation
Correct answer: A - Learns the training data including its noise too well and performs poorly on new, unseen data
Overfitting occurs when a model captures not only the genuine patterns but also the random noise and idiosyncrasies of the training data, giving excellent training accuracy but poor generalisation to new data. The opposite problem, where a model is too simple to capture the pattern, is called underfitting. The other options describe unrelated resource constraints.
Q6
medium
Consider the following statements about machine learning:
1. In supervised learning, the model is trained on data that includes labelled outputs.
2. In unsupervised learning, the algorithm discovers patterns in data without predefined labels.
3. Overfitting occurs when a model performs well on training data but poorly on unseen data.
How many of the statements given above are correct?
AOnly one
BOnly two
CAll three
DNone
Show answer and explanation
Correct answer: C - All three
All three statements are correct. Supervised learning uses input-output pairs where outputs are labelled (statement 1). Unsupervised learning finds structure such as clusters in unlabelled data (statement 2). Overfitting is precisely the phenomenon where a model fits the training data, including its noise, too closely and generalises poorly to new data (statement 3). These are standard definitions in machine learning, so the answer is all three.
Q7
hard
In machine learning, which one of the following statements correctly distinguishes 'supervised learning' from 'unsupervised learning'?
ASupervised learning uses labelled training data, whereas unsupervised learning finds patterns in unlabelled data
BSupervised learning works only on images, whereas unsupervised learning works only on text
CSupervised learning does not require any data, whereas unsupervised learning requires large datasets
DSupervised learning always uses neural networks, whereas unsupervised learning never does
Show answer and explanation
Correct answer: A - Supervised learning uses labelled training data, whereas unsupervised learning finds patterns in unlabelled data
The defining distinction is that supervised learning is trained on labelled data (inputs paired with known correct outputs) so the model learns to predict the label, while unsupervised learning works on unlabelled data to discover structure such as clusters or associations; this makes the first option correct. The second option is false because both approaches apply to many data types, not exclusively images or text. The third is false because supervised learning very much requires data, indeed labelled data. The fourth is false because neural networks can be used in both paradigms, and supervised learning also uses many non-neural algorithms. Hence the first statement is the correct distinction.
Q8
medium
In machine learning, the term 'overfitting' best describes a situation where a model:
APerforms well on the training data but poorly on new, unseen data because it has learned noise and idiosyncrasies of the training set
BIs too simple to capture the underlying pattern, performing poorly on both training and test data
CConsumes excessive memory and cannot fit within the available hardware
DProduces identical outputs regardless of the input fed to it
Show answer and explanation
Correct answer: A - Performs well on the training data but poorly on new, unseen data because it has learned noise and idiosyncrasies of the training set
Overfitting occurs when a model captures not only the signal but also the random noise and specific quirks of the training data, so it generalises poorly to unseen data despite high training accuracy. The second option describes underfitting (high bias). Hardware memory limits and constant outputs are unrelated concepts.
Practice all 44 Artificial Intelligence & Robotics questions free
Timed practice, instant scoring, and explanations for every question. Free forever - no card, no catch.