Artificial Intelligence and Robotics

1.Artificial Intelligence and Machine Learning

• Meaning of Artificial Intelligence, the Turing Test and the Origin of the TermImportance 4/5
  What does artificial intelligence (AI) mean? Remember the definition. Who proposed the Turing Test, and who coined the term "Artificial Intelligence"? Understand what the test checks, and what passing it does and does not prove.
• Narrow AI, Artificial General Intelligence and SuperintelligenceImportance 3/5
  How do narrow AI, artificial general intelligence (AGI) and superintelligence differ? Understand what AGI would need, such as broad generalisation, common sense and transfer of learning to unfamiliar tasks. Does high benchmark performance by a model such as GPT-6 Astra, for example in coding or mathematics, prove AGI?
• AI, Machine Learning, Deep Learning and Data Science ComparedImportance 4/5UPSC 2025
  How do AI, machine learning (ML), deep learning (DL) and data science relate to each other? Know which one sits inside which, and what each does. Is data science strictly a part of AI, or does it only overlap with it?
• Supervised, Unsupervised and Reinforcement LearningImportance 4/5
  How do machines learn from labelled data, from unlabelled data and from rewards? Know one everyday example of each. How does training adjust a model to reduce its prediction error? Remember where reinforcement learning from human feedback (RLHF) is used in chatbots.
• Natural Language Processing and Computer VisionImportance 3/5
  What do natural language processing (NLP) and computer vision do? Understand how NLP supports speech recognition, translation and chatbots, and how computer vision supports object and face detection.
• Algorithmic Bias and Explainable AIImportance 3/5UPSC 2026
  Where does bias in AI come from? Understand why unrepresentative training data leads to unfair outcomes, and why many models are "black boxes" whose reasoning cannot be explained. What does explainable AI try to achieve? Are the outputs of large language models free of bias?
• Artificial Neural Networks and Deep LearningImportance 4/5
  How is an artificial neural network built from neurons, layers and weights, and what makes it "deep"? Who received the 2024 Nobel Prize in Physics for foundational work on neural networks?
• Neural Network Architectures: CNNs, RNNs and TransformersImportance 3/5
  What is each of these good at: convolutional neural networks (CNNs), recurrent neural networks (RNNs) and transformers? Understand why the transformer, which uses attention over a whole sequence, powers large language models. In which year was it introduced?
• Neuromorphic Computing and Spiking Neural NetworksImportance 3/5
  How do neuromorphic chips imitate the brain, and why do they save energy? Understand the difference between keeping memory and processing in one place and moving data between them. Remember IBM NorthPole and Intel Hala Point as examples of brain-inspired hardware. What are spiking neural networks, and which of the two chips uses them? How does neuromorphic AI differ from embodied AI?
• Synthetic Intelligence and Biological ComputersImportance 3/5
  What is synthetic intelligence (SI), and how does it differ from AI? Note that this is a loosely used label for computing with living neurons, not a settled technical term. Remember the DishBrain experiment and the CL1 biological computer. Understand the energy argument, with the human brain as the comparison, and the ethical questions such computers raise.

2.Generative AI and Large Language Models

• Generative AI and Its Model Types (GANs, Diffusion Models, Transformers)Importance 4/5
  How does generative AI differ from predictive AI? Understand how a generative adversarial network (GAN), a diffusion model and a transformer-based text model each produce new content. Know the kinds of output: text, images and video, with Varya as an Indian text-to-video example.
• How Large Language Models WorkImportance 5/5UPSC 2026
  How does a large language model (LLM) produce its answers? Understand tokens, next-token prediction by probability, and training that reduces error. Remember parameters, pre-training versus fine-tuning, and the context window. Do LLMs understand language the way humans do?
• Multimodal Models, Small Language Models and Model DistillationImportance 3/5
  What is a multimodal model, and what can it handle beyond text? How do small language models differ from LLMs in cost and on-device use? Understand distillation, in which a smaller "student" model learns from a larger "teacher" model.
• Open-Source, Open-Weight and Closed AI ModelsImportance 3/5
  What does each kind of release make public: open-source, open-weight and closed? Understand why local use and less dependence on the cloud matter for sovereignty, and how open release raises the risk of misuse.
• Agentic AI and AI AgentsImportance 4/5
  How does an AI agent differ from a chatbot that only answers? Understand how an agent plans and carries out tasks using tools, memory and a sandbox, and what multi-agent systems and computer use mean. Understand its main risks, such as unintended actions, loss of control and data exposure.
• Information Agents and Concerns About Autonomous AgentsImportance 2/5
  What do information agents do on a user's behalf? What concerns do they raise for personal data, for traffic on websites, for the income of publishers, and for responsibility when their advice causes harm?
• Hallucination in Generative AIImportance 3/5
  What is hallucination in generative AI? Understand why models produce fluent but fabricated output, why this is dangerous in high-stakes uses such as courts, and how grounding and human checking reduce it.
• Model Collapse and Synthetic Training DataImportance 4/5
  What is model collapse, and what causes it? Understand where synthetic data is useful and where it is limited. What did researchers recently find about how much real, human-made data is needed to prevent collapse, and in what kind of models was it shown?
• Dual-Use Frontier AI in Cybersecurity (Claude Mythos, Project Glasswing)Importance 2/5
  Why is an AI model that finds flaws in software, such as Claude Mythos, called dual-use? Understand why access to such a model is limited to vetted organisations under Project Glasswing.
• AI Chatbots and Child SafetyImportance 2/5
  What risks do AI chatbots pose to minors, such as harmful conversations and emotional dependence? What safeguards has OpenAI announced for its teen version of ChatGPT?

3.Applications and Impacts of AI

• Capabilities and Limits of Present-Day AIImportance 4/5UPSC 2020
  What kinds of task can present-day AI do well: prediction, generation, perception or physical-world tasks? Which tasks is it not designed for? Understand why it lacks real understanding and common sense.
• Economic, Social and Environmental Impacts of AIImportance 3/5
  How might AI affect jobs and inequality, and why is its effect on jobs uncertain? Understand the energy and water use of data centres, the spread of misinformation, and the digital divide between paying and free users of AI tools.
• AI in Healthcare (IndiaAI–ICMR MoU)Importance 4/5
  How is AI used in diagnosis, disease surveillance and medical research? What does the IndiaAI–ICMR MoU provide in terms of datasets, computing and joint solutions? Understand the privacy concerns over health data.
• AI in AgricultureImportance 3/5
  How do AI and satellite data help in mapping fields and monitoring crops during the season? Remember ALU and AMED, the two Google DeepMind models, and what each one does.
• AI for Indian Languages (Bhashini, SraVaani)Importance 4/5
  Why does speech and translation AI matter for India's many languages? Remember what Bhashini is and which ministry runs it. Who built SraVaani, what does it do, and how is it made available to the public?
• AI in Security and Surveillance (Prajna)Importance 3/5
  How does AI help to analyse satellite images for security? Which DRDO centre built Prajna, and which ministry received it? Understand its uses in border management, counter-terrorism and spotting unusual activity.
• AI in Scientific Discovery (AlphaFold)Importance 3/5
  How does AI predict the structure of proteins? Who shared the 2024 Nobel Prize in Chemistry for AlphaFold and for protein design? Understand how AI is speeding up drug and vaccine research.

4.Facial Recognition, Biometrics and Deepfakes

• Biometric Authentication and Its TypesImportance 4/5
  How do physiological biometrics (such as fingerprint, iris and face) differ from behavioural ones (such as voice, gait and typing)? What is the difference between 1:1 verification and 1:N identification? Understand liveness detection and spoofing.
• Facial Recognition TechnologyImportance 4/5
  How does facial recognition work, from detecting a face to mapping its features and matching it? Understand why it can give false matches and show bias across groups. What are the privacy issues, and how do the Puttaswamy privacy test and the DPDP Act, 2023 apply?
• Facial Recognition Systems in India (DigiYatra, Aadhaar Face Authentication, NAFRS)Importance 4/5
  What is each system used for, and who runs it? Remember how DigiYatra works at airports and where passenger data is kept. Is Aadhaar face authentication 1:1 or 1:N, and does it need consent? Which agency runs the National Automated Facial Recognition System (NAFRS)?
• Emerging Biometric Methods (VitalID)Importance 2/5
  How can a body's internal vibrations from breathing and heartbeat work as a biometric? Understand how extended reality (XR) headset sensors capture them, and why this allows continuous login without a password.
• Deepfakes: Creation, Detection and HarmsImportance 4/5
  How are deepfakes made using GANs and diffusion models? Understand how they are detected, through detection tools, watermarking and content provenance. What harms do they cause, such as fraud, misinformation and non-consensual images targeting women?
• Rules on Synthetically Generated Information (IT Rules Amendment, 2026)Importance 4/5
  How do the 2026 amendments to the IT Rules define "synthetically generated information"? Remember what platforms must now do about labelling and metadata, how takedown times changed, and what platforms risk losing if they fail. Are routine good-faith edits covered?

5.AI Governance in India

• IndiaAI Mission and Its Seven PillarsImportance 5/5
  Which ministry runs the IndiaAI Mission? Remember the year it was approved, its outlay and its implementing body. What are its seven pillars? Understand the idea of "making AI in India and making AI work for India". Which sectors have AI Centres of Excellence (CoEs), and which one was announced in Budget 2025?
• AIKosh, the IndiaAI Compute Portal and Indigenous Foundation ModelsImportance 4/5
  What do AIKosh and the compute portal provide to researchers and startups? How many GPUs are now available under the Mission? Remember BharatGen and the startups chosen to build India's own foundation models, and what was launched at the India AI Impact Summit 2026.
• AI Skilling Initiatives (IndiaAI FutureSkills, YUVA AI for All, Agentic AI Skilling Initiative)Importance 2/5
  Which bodies run IndiaAI FutureSkills, YUVA AI for All and the Agentic AI Skilling Initiative? Remember the Data and AI Labs in smaller cities, and the two programmes under the Agentic AI initiative, run by NIELIT with Intel.
• India AI Governance Guidelines, 2025Importance 5/5
  Who issued the India AI Governance Guidelines, 2025? Remember the seven sutras and the six pillars. What roles do the AI Governance Group (AIGG), the Technology and Policy Expert Committee (TPEC) and the AI Safety Institute (AISI) play? Why do the Guidelines rely on existing laws and sector regulators instead of a separate AI law?
• IndiaAI Safety InstituteImportance 3/5
  Why was the IndiaAI Safety Institute set up, and under which pillar of the Mission? Understand its hub-and-spoke model and its work on research, standards and testing. Is it a regulator?
• NITI Aayog's National Strategy for AI and Responsible AI PrinciplesImportance 3/5
  What was the #AIforAll strategy, and in which year did NITI Aayog bring it out? Remember its five focus sectors. What do the Responsible AI principles papers cover?
• RBI's FREE-AI FrameworkImportance 3/5
  What does FREE-AI stand for? Remember when the RBI released it, and its seven sutras, six pillars and 26 recommendations. How did the national Guidelines use its sutras?
• Supreme Court's Draft Regulations for Use of AI in Courts, 2026Importance 4/5
  Which uses of AI do the draft regulations permit, and which do they bar? Understand the reasons: bias, lack of transparency, hallucination and accountability. Must lawyers disclose their use of AI? Have the regulations been finalised?
• AI Tools and Policies in Indian Courts (SUPACE, SUVAS, Kerala High Court Policy)Importance 2/5
  What do the court tools SUPACE and SUVAS do? Remember what the Kerala High Court's 2025 policy for the district judiciary says about using AI in decisions.

6.Global AI Governance

• India AI Impact Summit 2026 and the New Delhi DeclarationImportance 5/5
  When and where was the India AI Impact Summit held? Remember the Declaration's guiding principle, its three sutras (People, Planet, Progress), its seven chakras and the Charter for the Democratic Diffusion of AI. What shift from "safety" to "impact" does the summit series show?
• Global AI Summit Series (Bletchley Park, Seoul, Paris)Importance 3/5
  In what order were the global AI summits held, and where? Remember the Bletchley Declaration, and what the Paris AI Action Summit added. Which two countries co-chaired Paris?
• EU AI Act and Risk-Based RegulationImportance 4/5
  What are the four risk tiers of the EU AI Act, with an example of each? Remember when the Act came into force, when the bans began, and when the rules for general-purpose AI began. Have the deadlines for high-risk systems been postponed, and has that change been formally adopted?
• UN Initiatives on AI Governance (Scientific Panel, Global Dialogue)Importance 3/5
  Which 2025 UN resolution created the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance? What did the panel warn about, including agentic AI? Where was the first Dialogue held, and where is the next?
• Other International AI Frameworks (GPAI, OECD Principles, UNESCO Recommendation, Council of Europe Convention)Importance 3/5
  What is each of these frameworks, and which of them binds countries legally? What is India's role in the Global Partnership on AI (GPAI)? Is India a party to the Council of Europe Convention?
• Lethal Autonomous Weapons SystemsImportance 2/5
  What are lethal autonomous weapons systems? Understand the "meaningful human control" debate. Which UN forum discusses them?

7.AI and Copyright

• Authorship of AI-Generated Works (DABUS Order, 2026)Importance 4/5
  What did the Registrar of Copyrights decide in 2026 on Stephen Thaler's DABUS application? How do originality (Section 13), authorship (Section 2(d)(vi)) and ownership (Sections 17–19) differ under the Copyright Act, 1957? Can an AI be the author of a work, and can that work still be original? Know the earlier RAGHAV and Suryast episode, and how the USA, the UK and China treat AI-generated works.
• Training AI on Copyrighted Works (ANI v OpenAI, DPIIT Working Paper)Importance 4/5
  Is training an AI model on copyrighted content infringement? Understand the fair dealing argument and what the Delhi High Court decided at the interim stage in ANI v OpenAI. Is the main suit still pending, and has a Division Bench decided ANI's appeal against the interim order? What hybrid licensing model has DPIIT proposed, and does it allow rights holders to opt out? Training and outputs raise separate questions.

8.Robotics

• Robotics: Meaning and Basic ComponentsImportance 3/5
  What is a robot? Understand the job of its sensors, actuators, controller and power source, and the sense, think and act loop. Machine vision and robotic arms are examples of parts. Asimov's laws are a fictional label, not law.
• Types of Robots (Industrial, Service, Humanoid, Collaborative, Swarm)Importance 3/5
  How do industrial, service, humanoid, collaborative (cobot) and swarm robots differ? Know a use and an example for each. Which kinds of robot are common platforms for embodied AI?
• Embodied AIImportance 3/5
  What is embodied AI? Understand how a robot's brain, body and surroundings together shape its learning, and what morphological computation means. Remember its uses and the simulation-to-real gap.
• Robots for Hazardous Work and Ending Manual Scavenging (G-SPIDER, Bandicoot)Importance 3/5
  What do G-SPIDER and Bandicoot do, and which company made them? Remember where G-SPIDER was deployed and under which scheme. How do they support the Act of 2013 against manual scavenging and the NAMASTE scheme?
• Vyommitra and Robots in Space MissionsImportance 3/5
  What is Vyommitra, and why is a half-humanoid robot flown before astronauts? Understand what it tests. Has the uncrewed Gaganyaan G1 mission that carries it been launched?
• India's Robotics Strategy and Institutions (Draft National Strategy on Robotics, CAIR)Importance 2/5
  What did MeitY's 2023 draft National Strategy on Robotics propose, including the proposed National Robotics Mission? Has it been finalised? Remember the roles of DRDO's CAIR and ARTPARK.

1. 1.

   Consider the following statements about the IndiaAI Mission:
   1. It was approved by the Union Cabinet in March 2024.
   2. Its outlay is ₹10,371.92 crore over five years.
   3. It is built on five pillars.
   4. Its vision is "Making AI in India and Making AI Work for India".
   How many of the above statements are correct?

   1. AOnly one
   2. BOnly two
   3. COnly three
   4. DAll four
2. 2.

   With reference to the India AI Governance Guidelines, 2025, consider the following statements:
   1. They conclude that a separate law to regulate AI is not needed at this stage.
   2. The AI Governance Group (AIGG) is chaired by the Principal Scientific Adviser.
   3. The AI Safety Institute has been given an enforcement role to penalise violators.
   Which of the statements given above is/are correct?

   1. A1 only
   2. B3 only
   3. C2 and 3 only
   4. D1 and 2 only
3. 3.

   Consider the following statements about the New Delhi Declaration adopted at the India AI Impact Summit, 2026:
   1. It is guided by the principle "Sarvajan Hitaya, Sarvajan Sukhaya".
   2. Its three sutras are People, Planet and Progress.
   3. It is organised around five chakras (pillars).
   How many of the above statements are correct?

   1. AOnly one
   2. BOnly two
   3. CAll three
   4. DNone
4. 4.

   With reference to large language models (LLMs), consider the following statements:
   1. Text is converted into tokens, which are small units of words or subwords, so that machines can process it.
   2. During supervised fine-tuning, the model is trained on curated prompt-response pairs created by human experts.
   3. An LLM produces its answer by retrieving a stored ready-made answer rather than by predicting the next word.
   Which of the statements given above is/are correct?

   1. A1 and 2 only
   2. B2 and 3 only
   3. C1 and 3 only
   4. D1, 2 and 3
5. 5.

   Consider the following pairs regarding the Nobel Prize in Chemistry, 2024:

   Laureate	Recognised for
   1. David Baker	Computational protein design
   2. Demis Hassabis and John Jumper	Development of AlphaFold2
   3. Demis Hassabis and John Jumper	Computational protein design

   Which of the pairs given above is/are correctly matched?

   1. A1 only
   2. B2 only
   3. C1 and 2 only
   4. D1, 2 and 3
6. 6.

   The European Union's AI Act was amended by the AI Omnibus, which entered into force on 27 July 2026. As of October 2026, consider the following statements about the Act's timeline:
   1. It entered into force on 1 August 2024.
   2. The rules on prohibited AI practices and AI literacy have applied since 2 February 2025.
   3. The obligations for high-risk use cases listed in Annex III still apply from 2 August 2026, the date originally scheduled.
   4. The governance rules and the obligations for general-purpose AI (GPAI) models became applicable on 2 August 2025.
   How many of the above statements are correct?

   1. AOnly one
   2. BOnly two
   3. COnly three
   4. DAll four
7. 7.

   Match List I with List II:

   List I (System / platform)	List II (Description)
   A. DigiYatra	1. Implemented by the National Crime Records Bureau to help identify persons
   B. National Automated Facial Recognition System	2. Consent-based 1:1 matching against the Aadhaar holder's enrolled facial image
   C. Aadhaar face authentication	3. Contactless processing of airport passengers based on facial recognition
   D. Bhashini	4. National public digital platform for languages, developed by MeitY

   Select the correct answer using the code given below:

   1. AA-1, B-2, C-3, D-4
   2. BA-2, B-1, C-4, D-3
   3. CA-3, B-2, C-1, D-4
   4. DA-3, B-1, C-2, D-4
8. 8.

   With reference to the global AI summit series, consider the following statements:
   1. The Paris AI Action Summit (2025) was the third summit in the series.
   2. India was one of the co-chairs of the Paris AI Action Summit.
   3. The Bletchley Park Summit (2023) was held after the Seoul Summit (2024).
   Which of the statements given above is/are correct?

   1. A1 and 2 only
   2. B2 and 3 only
   3. C3 only
   4. D1, 2 and 3
9. 9.

   Match List I with List II:

   List I (Summit)	List II (Outcome)
   A. Bletchley Park (2023)	1. Proposed a network of AI Safety Institutes
   B. Seoul (2024)	2. Declaration advocating safe, human-centric and responsible AI
   C. Paris (2025)	3. Joint Statement on Inclusive and Sustainable AI, not signed by the US and the UK
   D. New Delhi (2026)	4. Declaration built on three sutras: People, Planet and Progress

   Select the correct answer using the code given below:

   1. AA-1, B-2, C-3, D-4
   2. BA-2, B-1, C-3, D-4
   3. CA-2, B-1, C-4, D-3
   4. DA-2, B-3, C-1, D-4
10. 10.

    As of February 2026, consider the following statements about the 2026 amendment to the IT Rules, 2021 on AI-generated content:
    1. The amendment was notified on 10 February 2026 and came into force on 20 February 2026.
    2. Content flagged by a court or the government must be taken down within three hours.
    3. Platforms may allow AI labels or metadata to be removed if a user requests it.
    Which of the statements given above is/are correct?

    1. A1 only
    2. B3 only
    3. C2 and 3 only
    4. D1 and 2 only
11. 11.

    Consider the following statements:
    Statement-I: Training an AI system mainly on AI-generated data can degrade its outputs, making them inaccurate, repetitive or biased.
    Statement-II: Model collapse is caused by excessive use of synthetic data, recycling of errors and biases in training, and a decline in fresh human-generated data.
    Which one of the following is correct in respect of the above statements?

    1. ABoth Statement-I and Statement-II are correct and Statement-II explains Statement-I
    2. BBoth Statement-I and Statement-II are correct and Statement-II does not explain Statement-I
    3. CStatement-I is correct but Statement-II is incorrect
    4. DStatement-I is incorrect but Statement-II is correct
12. 12.

    Which one of the following best distinguishes an AI agent from a traditional chatbot?

    1. AIt can run only on a dedicated supercomputer and not through ordinary apps
    2. BIt is trained exclusively on synthetic data without any human-written text
    3. CIt plans and completes tasks on its own instead of only answering questions
    4. DIt processes text only and has no internet access or memory of past tasks
13. 13.

    Consider the following pairs:

    System / model	Description
    1. Prajna	AI-enabled satellite imaging system developed by DRDO's Centre for Artificial Intelligence and Robotics, received by the Ministry of Home Affairs
    2. SraVaani	Multilingual speech recognition model developed by IISc's SPIRE Lab with ARTPARK and Google
    3. ALU (Agricultural Landscape Understanding)	AI model that tracks crop types, sowing and harvesting

    Which of the pairs given above is/are correctly matched?

    1. A1 only
    2. B1 and 2 only
    3. C2 and 3 only
    4. D1, 2 and 3
14. 14.

    Consider the following statements about the MoU between IndiaAI and the Indian Council of Medical Research (ICMR) on AI in healthcare:
    1. ICMR will provide anonymised, ethics-approved health datasets developed under the MIDAS framework to the AIKosh platform.
    2. ICMR will provide GPU-based computing infrastructure at subsidised rates.
    3. IndiaAI is a mission under the Ministry of Health and Family Welfare.
    4. The two bodies will jointly develop AI-driven solutions for disease surveillance and diagnostics.
    How many of the above statements are correct?

    1. AOnly one
    2. BOnly two
    3. COnly three
    4. DAll four
15. 15.

    Consider the following statements:
    1. Machine learning is a subset of artificial intelligence.
    2. Deep learning is a subset of machine learning.
    3. Artificial intelligence is a subset of machine learning.
    Which of the statements given above is/are correct?

    1. A1 and 2 only
    2. B2 and 3 only
    3. C1 and 3 only
    4. D1, 2 and 3
16. 16.

    With reference to copyright and AI-generated works in India, consider the following statements:
    1. The Copyright Act, 1957 does not define "originality" in Section 13.
    2. In Eastern Book Company v. D.B. Modak, the Supreme Court held that a work need not be novel or inventive but must show a minimum degree of creativity.
    3. In the DABUS matter, the Copyright Office held that an AI system can presently be recognised as the statutory author of a work.
    Which of the statements given above is/are correct?

    1. A1 only
    2. B1 and 2 only
    3. C2 and 3 only
    4. D1, 2 and 3
17. 17.

    As of June 2026, which one of the following correctly describes the Supreme Court's Draft 'Regulations for Use of Artificial Intelligence (AI) in Courts, 2026'?

    1. AThey permit AI in adjudicatory functions if its use is disclosed, but bar it for research
    2. BThey bar every use of AI in courts, including administrative work such as listing
    3. CThey allow AI to draft judgments if the presiding judge discloses its use
    4. DThey allow AI for administrative and research work but not for adjudication
18. 18.

    With reference to the Automated Facial Recognition System in India, which one of the following is a concern raised by critics?

    1. AIt records only the fingerprints of arrested persons and cannot match CCTV images
    2. BIt is operated by a private company under licence, outside the control of the police
    3. CIt returns definitive matches that leave no room for error or human review
    4. DIt lacks a statutory basis and gives only probabilistic matches
19. 19.

    Consider the following statements:
    Statement-I: An open-weight AI model is not necessarily an open-source model.
    Statement-II: An open-weight model lets its parameters be downloaded, run and customised locally, but does not require the developer to disclose code or training methods.
    Which one of the following is correct in respect of the above statements?

    1. ABoth Statement-I and Statement-II are correct and Statement-II explains Statement-I
    2. BBoth Statement-I and Statement-II are correct and Statement-II does not explain Statement-I
    3. CStatement-I is correct but Statement-II is incorrect
    4. DStatement-I is incorrect but Statement-II is correct
20. 20.

    Which one of the following was launched on 2 June 2025 as India's first government-funded multimodal large language model, supporting 22 languages?

    1. ABhashini
    2. BAIKosh
    3. CBharatGen
    4. DSraVaani

1. 1.

   UPSC 2026

   Which of the following statements with regard to Large Language Models (LLMs) used in machine learning is/are correct ?
   1. LLMs assign probabilities to the next possible words and then pick the one with the highest probability.
   2. LLMs process data through mathematical optimization to minimise prediction errors.
   3. LLMs produce unbiased outputs.
   Select the answer using the code given below :

   1. A1 only
   2. B1 and 2 only
   3. C2 and 3 only
   4. D1, 2 and 3
2. 2.

   UPSC 2020

   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 :

   1. A1, 2, 3 and 5 only
   2. B1, 3 and 4 only
   3. C2, 4 and 5 only
   4. D1, 2, 3, 4 and 5
3. 3.

   UPSC 2025

   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?

   1. AI and II only
   2. BII and III only
   3. CI and III only
   4. DI, II and III