Computing and AI in UPSC Prelims: 16 questions that reward a clear definition
UPSC asked about computing, AI and blockchain in 7 of the last 11 Prelims papers. What those 16 questions test, why a half-remembered definition is the usual trap, and a plan built on them.
Most aspirants prepare technology from the news: a new chip, a new app, a new buzzword. They learn the name and move on. The past papers suggest the name is not enough. UPSC usually hands you the name. What it tests is whether you know what the term means, well enough to spot a definition that has been bent or swapped.
This post looks at the 16 questions on computing, AI and blockchain asked from 2016 to 2026 and turns them into a plan.
What the numbers say
Three things stand out.
It arrived in 2018 and peaked in 2022. There were no questions in 2016 or 2017. Since then, 7 of the 9 papers had at least one. The 2022 paper alone had 5, four of them almost back to back as questions 31, 32, 33 and 35.
| Year | Questions |
|---|---|
| 2016–2017 | 0 |
| 2018 | 3 |
| 2019 | 1 |
| 2020 | 2 |
| 2021 | 0 |
| 2022 | 5 |
| 2023 | 0 |
| 2024 | 1 |
| 2025 | 1 |
| 2026 | 3 |
It is mostly a list to judge. 13 of the 16 questions asked you to judge a list: 10 statement sets, one set of pairs, and two lists of items (what AI can do, and which apps are built on open source). Only 3 were single-answer questions, and every question since 2025 has been a statement set.
Two families keep coming back. By our count, 5 questions touch blockchain: a pair in 2018, a statement set in 2020, Web 3.0 and NFTs in 2022, and blockchain features in 2026. Quantum computing had 3: the qubit in 2022, the Majorana 1 chip in 2025 and the National Quantum Mission in 2026.
What UPSC is really testing
By our count, the 16 questions fall into four groups:
- What the term means: 9 questions. The Internet of Things, AR and VR, blockchain (twice), Web 3.0, Software as a Service, NFTs, the metaverse and Large Language Models.
- Which field a term belongs to: 2. Matching Belle II, blockchain and CRISPR-Cas9 to their subjects in 2018, and placing the word “qubit” in 2022.
- India’s own platforms and missions: 3. Aadhaar’s open APIs, whether Aarogya Setu, CoWIN, DigiLocker and DIKSHA are built on open-source platforms, and the National Quantum Mission.
- Where the technology stands today: 2. What AI could effectively do in 2020, and the Majorana 1 quantum chip in 2025.
So 11 of the 16 turned on knowing what a term means or where it belongs. Even the single-answer questions worked this way: two gave a description and asked for the term (the Internet of Things, the metaverse), and one gave the term and asked for its field (the qubit).
The trap most questions share
The wrong option is a definition that is almost right. UPSC swaps two definitions, borrows from a technology’s most famous use, or offers a neighbour that sounds just as technical. If you only half know the term, the bent version reads as true.
Here is where the trap came from in six of them:
| Term (year) | What it actually means | What the trap borrowed |
|---|---|---|
| AR and VR (2019) | AR adds to the real world; VR shuts it out | Each other’s definition |
| Blockchain (2020) | A shared ledger for many kinds of data | Bitcoin, its most famous use |
| NFTs (2022) | Unique tokens on a blockchain | Money, because NFTs are bought and sold |
| Qubit (2022) | The basic unit of quantum computing | Cloud services, which can offer quantum computers |
| Metaverse (2024) | Connected 3D virtual worlds | “Virtual matrix”, which sounds right |
| Majorana 1 (2025) | A quantum chip from Microsoft | AWS, which announced a different chip |
The 2018 pairs question mismatched whole subjects. It linked Belle II, a particle physics experiment in Japan, to artificial intelligence, and CRISPR-Cas9, a gene-editing tool, to particle physics. Only the blockchain pair was correct. In 2020, wireless transmission of electricity was offered as an AI task. It sounds “smart”, but it is physics.
The list questions have a second trap, and it runs the other way. Leaving aside 2026, whose answers are still provisional, 4 of the 10 had every item true: Aadhaar’s open APIs (2018), the open-source apps, Web 3.0 and SaaS (all 2022). The 2022 paper set three of these in a row, as questions 31, 32 and 33.
Three of them had an item that sounded too good. Web 3.0 is “operated by users collectively rather than a corporation”, which sounds idealistic. It is the defining idea of Web 3.0. SaaS buyers “can customise the user interface”, which sounds unlikely for a packaged service. Customising within the provider’s settings is common. Government apps sound closed, but all four were built on open-source platforms.
How to prepare
1. Write a one-line definition for each term, in your own words. IoT: everyday devices connected over the internet that act on their own. Qubit: the basic unit of quantum computing. Metaverse: connected 3D virtual worlds that many users share.
2. Learn the terms in pairs. AR and VR. Fungible and non-fungible. Bit and qubit. Machine learning and deep learning: deep learning is a subset of machine learning, which is a subset of AI. Pairs are where swapped definitions come from.
3. Separate a technology from its most famous use. Blockchain is not only cryptocurrency. NFTs are traded, but they do not work as money.
4. Don’t drop the idealistic statement. For Web 3.0, the statement that sounded too good was the definition itself.
5. For a product in the news, know who made it and what it is for. Majorana 1 is Microsoft’s quantum chip, meant to enable scalable quantum computing. The trap named another company, AWS, as its maker.
6. Keep a short list of India’s digital platforms. For Aadhaar, DigiLocker, CoWIN and DIKSHA, note what each does and whether it is built on open technology.
7. Solve the past questions before reading more. Start with the computing, AI and blockchain questions, then see the wider IT, electronics and emerging tech topic.
What to skip: programming, the history of computing and how a chip is built. None of the 16 questions needed them.
Two questions that show the pattern
2019 · Swapped definitions. Four statements on Augmented Reality and Virtual Reality. Statements 1 and 2 swap the two definitions: they say AR shuts out the real world and VR projects images onto it. It is the other way round. Statements 3 and 4 have them right, so the answer is 3 and 4. See the question
2022 · The famous neighbour. Three statements on Non-Fungible Tokens. NFTs can represent physical assets, and each is a unique token on a blockchain. But they cannot be “exchanged at equivalency” like money, because non-fungible means not interchangeable. The trap was that NFTs are bought and sold, which makes them feel like currency. See the question
For the other high-yield areas, see the must-do list. Biotechnology questions also test what a technique does rather than its name, and we wrote about them here.