Which one of the following is the characteristic of Artificial Intelligence?

Correct — A, Replicates human decision making. Artificial intelligence is the branch of computer science concerned with building systems that perform tasks which would ordinarily require human intelligence — reasoning about a situation, learning from experience, recognising patterns in images or speech, understanding language, and arriving at a judgment. Of those, judgment is the one that distinguishes the field from every other kind of software, and it is what the option captures: an artificial-intelligence system is expected to produce a decision on new input rather than simply retrieve a stored answer. That is why a spam filter that classifies a message it has never seen, a diagnostic system that reads a scan, and a credit model that scores an applicant are all described as artificial intelligence, while a program that looks up a value in a table is not, however large the table. The other three options each describe a capability that ordinary software has had for decades. Storing information and organising data for a purpose are what a database management system does, and storage is a precondition of intelligence rather than an instance of it — the data has to exist before a model can be trained on it, but holding it is not deciding anything. Allowing a user to interact with media is multimedia, the subject of the question immediately before this one in the same paper. One honest qualification is worth carrying: these systems approximate the outcomes of human judgment by statistical means, and do not reason in the way a person does, so 'replicates' should be read as replicating the function rather than the mechanism.

• (b)Stores relevant information — That is storage, which a database or a file system provides. Holding information is a precondition for training a model, not a characteristic that distinguishes artificial intelligence from any other software.
• (c)Stores similar kind of data for a specific purpose — A description of a database or a data warehouse — structured storage organised for a defined purpose. It says nothing about inference, learning or judgment.
• (d)Allows user to interact with media — That is interactive multimedia, which the previous question in this very paper defines. Interactivity concerns how content is presented, not whether the system reasons about anything.

Artificial intelligence covers techniques by which machines carry out tasks associated with human intelligence. Its dominant approach today is machine learning, where a system is not programmed with explicit rules but learns a mapping from data — supervised learning from labelled examples, unsupervised learning from structure in unlabelled data, and reinforcement learning from reward and penalty. Deep learning, using artificial neural networks with many layers, drives most recent progress in computer vision, speech recognition and language. Older symbolic approaches, notably rule-based expert systems, encoded human expertise directly and remain in use where decisions must be explainable.

Questions in this family test whether you can separate intelligence from mere capacity. Storage, retrieval, speed and interactivity are things computers have always had; what artificial intelligence adds is inference on inputs the system has not seen before. Read each option and ask whether an ordinary database or media player could do it — if yes, it is not what makes a system intelligent. The same test explains the UPSC 2020 item on what artificial intelligence can effectively do, whose official key credited reducing industrial power consumption, diagnosing disease and converting text to speech, and rejected wireless transmission of electrical energy, which is a physics problem no amount of intelligence solves. Anchor the field's status to the paper's own date: at the April 2025 examination, large language models were in everyday public use, India's national mission on artificial intelligence was running, and the debate had moved to regulation, model safety and the electricity and water demands of data centres. Those developments are all after the syllabus's older material, so state the general definition first and the current example second.

• Artificial intelligence is the building of systems that perform tasks normally requiring human intelligence — reasoning, learning, perception, language and decision-making.
• Machine learning is its dominant branch, learning a mapping from data rather than from explicit rules.
• Learning is described as supervised when examples are labelled, unsupervised when they are not, and reinforcement when the system learns from reward and penalty.
• Deep learning uses artificial neural networks with many layers and underlies most recent progress in vision, speech and language.
• Expert systems encode human expertise as explicit rules and are the older, symbolic approach.
• Storage and retrieval belong to databases; interactivity with media belongs to multimedia — neither is a characteristic of artificial intelligence.
• These systems approximate the outcomes of human judgment statistically; they do not reason by the same mechanism a person uses.

• Equating large data storage with intelligence; a database that holds everything decides nothing.
• Treating interactivity as intelligence — a responsive interface is multimedia, not inference.
• Assuming machine learning and artificial intelligence are the same thing; machine learning is one branch of a wider field that also includes symbolic and rule-based methods.
• Reading 'replicates human decision making' as a claim that the machine thinks like a person; it reproduces the function, not the mechanism.

As a characteristic-or-definition item like this one, as a list of tasks with the question of which artificial intelligence can effectively perform, or through the nesting of artificial intelligence, machine learning and deep learning.

• practice — not a real PYQ

  Machine learning is best described as which one of the following?

  • (a)A branch of artificial intelligence in which systems learn patterns from data rather than from explicit rules
  • (b)A method of storing data in tables with rows and columns
  • (c)A technique for compressing image files
  • (d)A protocol for transferring files over the internet

  Answer(a) A branch of artificial intelligence in which systems learn patterns from data rather than from explicit rules — deep learning is in turn a branch of machine learning.

• practice — not a real PYQ

  A system in which human expertise is encoded as a set of explicit rules to reach conclusions is known as which one of the following?

  • (a)Expert system
  • (b)Operating system
  • (c)Database management system
  • (d)File system

  Answer(a) Expert system — the older, symbolic approach to artificial intelligence, valued where a decision has to be explainable.