← Explore exam analytics

UPSC CSE · Prelims GS-I · Science and technology · TOPIC LESSON

Technology and applications

Understand technology through mechanisms, measurement boundaries and evidence rather than labels.

UPSC Civil Services115 min study guide10 indexed sections

BY THE END, YOU CAN

  • Explain semiconductor control and energy-conversion pathways.
  • Separate stored energy, output power and conversion efficiency.
  • Distinguish digital infrastructure, permissions and protective functions.
  • Interpret AI metrics and satellite sensing under stated conditions.
  • Evaluate biotechnology and nanomaterial applications with lifecycle safeguards.

BEFORE YOU START

  • Basic electricity, energy, waves and cell/genetic concepts.
  • Ratios, percentages and powers of ten.

Original educational lesson. Work through the examples before revealing the practice answers. This original common-core lesson supports its declared paper scope. Optional subjects, qualifying language papers and the interview are outside this resource.

LEARN THROUGH ORIGINAL PRACTICE

Original UPSC-style worked applications and practice

All teaching examples, cases, prompts and model answers are original. Official exam questions and answer keys are not reproduced. This common-core lesson does not imply complete historical paper coverage or an official score.

01 / UNDERSTAND & APPLY

From mechanism to a justified application

Begin a technology claim with its mechanism: what physical, chemical or biological process produces the intended result? Then identify the useful output, comparator and operating conditions. A sensor detecting a substance in a controlled laboratory is different from a reliable field service that measures it accurately, communicates the result and supports a sound decision. Deployment also needs maintenance, skills, supplies and institutions. A successful demonstration does not by itself establish affordability or large-scale reliability.

Efficiency compares compatible useful output and input. Capacity describes a rated capability under stated conditions, while actual output depends on operation. Reliability concerns whether the required service is delivered consistently, and cost includes a specified set of expenditures. An efficient component can belong to a poorly performing whole system. Compare technologies at the same functional boundary, such as delivered illumination of the same quality for the same period, rather than comparing a device rating with a service outcome.

Lifecycle assessment broadens the account to manufacture, operation, maintenance and end-of-life. A device with no direct exhaust can still require energy and materials in production and disposal. Conversely, an initial material cost may be offset by lower operational use over a long life. Neither conclusion follows from the label clean or digital alone. Identify the boundary, expected lifetime and alternatives before claiming an overall benefit.

02 / UNDERSTAND & APPLY

Semiconductors and electronic control

Semiconductors have electronic properties that can be deliberately controlled. Doping introduces selected impurities that alter the availability of mobile electrons or holes. A hole represents an unoccupied electronic state that can behave as a positive charge carrier, rather than a literal cavity in the material. Carrier populations and mobility affect conductivity, and conditions such as temperature matter. Doping changes a material’s response; it does not create unlimited energy.

In a field-effect transistor, the electric field associated with a gate controls conduction through a channel. That control allows switching between comparatively conducting and non-conducting states or amplification within an appropriate circuit. Large combinations of devices implement logic and memory. A transistor’s energy comes from the circuit supply, so switching is not energy generation. A piece of semiconductor is also not automatically a complete computer or a functioning solar cell; device structure is essential.

A diode has an asymmetric current–voltage response over its operating range; the ideal one-way-valve description has limits, including leakage and breakdown. Light-emitting diodes convert electrical input into light through electronic transitions, while photovoltaic devices convert suitable absorbed light into electrical output. The directions of conversion differ, and neither conversion is perfectly efficient. Material selection, device design and conditions determine performance.

ORIGINAL PRACTICE · 1

Does doping create unlimited electrical energy inside a transistor?

Reveal answer & reasoning

No.

It alters charge-carrier availability; operation still obeys energy conservation and requires a circuit energy source.

03 / UNDERSTAND & APPLY

Solar power, batteries and hydrogen

In a photovoltaic device, absorbed light can create mobile charge carriers; the device structure separates and collects them through an external circuit. Conversion efficiency compares electrical output with incident radiant input over the same area and interval. For the fictional 2 m² module at 800 W/m², incident power is 1600 W and electrical output 320 W gives 20% efficiency. Over two hours, the respective energies are 3.2 kWh and 0.64 kWh. These are module-boundary figures, not annual yield or electricity delivered after downstream losses.

A rechargeable battery stores energy through electrochemical changes that can be reversed to a useful extent during charging. During discharge, coupled reactions and ion/electron transport support electrical output. Energy capacity in kWh differs from power capability in kW: a large store is not necessarily able to discharge at any requested rate. The fictional store starts fully charged with 10 kWh of stored energy at its model rating and allows 80% use, leaving 8 kWh internally available; a 90%-efficient output converter delivers 7.2 kWh. At a constant 1.8 kW load, the supplied model gives four hours.

A fuel cell continuously converts the chemical energy of supplied reactants into electrical energy and heat while appropriate supplies and conditions are maintained. In a hydrogen cell, the overall reaction with oxygen produces water. In a proton-exchange design, electrons take an external circuit while protons cross the electrolyte, illustrating how electrochemical separation supports useful current. Other cell types differ. Water at the point of operation does not erase emissions from hydrogen production, electricity supply, transport or equipment manufacture.

Hydrogen is an energy carrier, and storage shifts energy availability across time. Neither creates energy from nothing. Round-trip efficiency divides recovered output by charging input across the same defined boundary: 40 kWh recovered after 50 kWh supplied is 80%. It differs from a single converter efficiency or an allowed usable fraction. The native figure keeps the PV and battery examples explicitly separate; it does not imply the 0.64 kWh PV output charged the independent 10 kWh store.

Full-size scientific diagram. Swipe or scroll horizontally; keyboard users can focus the diagram and use arrow keys.

Separate photovoltaic and storage accountsTwo independently bounded original examples. PV:2m² at800W/m² receives1600W; over2h incident3.2kWh, electrical320W/0.64kWh, efficiency20%. Separate battery: initially fully charged with10kWh stored energy at its model rating,80%allowed use gives8kWh,90%converter gives7.2kWh delivered; at1.8kW lasts4h. No link between PV output and battery input. All values/assumptions in prose.Energy accounts: two separate fictional systemsValues use the same boundary within each example. No energy link is asserted between A and B.A. PV module over two hoursIncident light: 3.2 kWh2 m² × 800 W/m² × 2 h20% efficiencyElectrical output: 0.64 kWh320 W × 2 hThe remaining input is not electrical output; no downstream or annual-yield claim.B. Independent battery and converter exampleNominal: 10 kWhInitially fully chargedUsable: 8 kWh80% allowed useDelivered: 7.2 kWh90% converterDelivered duration = 7.2 kWh / 1.8 kW = 4 hours under supplied assumptions.Energy capacity (kWh) does not specify maximum discharge power (kW).
The PV and battery examples are separate. Capacity, useful energy and power use different denominators.

ORIGINAL WORKED EXAMPLE

Photovoltaic input and output

A fictional PV module of area 2 m² receives steady irradiance 800 W/m² and supplies 320 W electrical output during a two-hour interval. Ignore changes during that interval.

  1. Incident power=area×irradiance=2×800=1600 W. Efficiency=320/1600=0.20=20%.
  2. Over two hours, electrical energy=0.320 kW×2 h=0.64 kWh and incident energy=1.6 kW×2 h=3.2 kWh. The energy ratio gives the same 20%.
  3. The supplied conditions are steady and the boundary is the module. No annual generation or downstream delivered-energy claim follows from this short interval.

Answer 20% conversion efficiency and 0.64 kWh module output.

ORIGINAL WORKED EXAMPLE

From nominal storage to delivered service

A storage unit starts fully charged with 10 kWh of stored energy at its nominal model capacity, allows use of 80% of that energy and delivers it through a 90%-efficient converter to a constant 1.8 kW load; ignore other losses and changing power limits.

  1. Initial stored energy is 10 kWh because the unit starts fully charged at its supplied model rating. The allowed usable fraction gives 10×0.80=8 kWh internally available.
  2. Converter output is 8×0.90=7.2 kWh. A constant 1.8 kW load lasts 7.2/1.8=4 hours under the stated no-other-loss/power-limit assumptions.
  3. The 80% usable fraction and 90% conversion efficiency describe different constraints. The 10 kWh capacity alone does not establish any maximum permitted discharge power.

Answer 7.2 kWh delivered and four hours under supplied assumptions.

ORIGINAL PRACTICE · 1

A store receives 50 kWh and later returns 40 kWh across the same measurement boundary. What is round-trip efficiency?

Reveal answer & reasoning

80%.

40/50×100; the 10 kWh difference is not new usable energy.

ORIGINAL PRACTICE · 2

Does water at a hydrogen fuel cell exhaust prove zero lifecycle greenhouse emissions?

Reveal answer & reasoning

No.

Hydrogen production, energy supply, equipment and transport belong in the lifecycle boundary.

04 / UNDERSTAND & APPLY

Data, networks and connected devices

Digital information is represented using bits. In the stated convention, eight bits form one byte, and decimal MB means one million bytes. File size and transmission rate must use compatible units: 120 MB is 960 Mbit, so a sustained payload throughput of 24 Mbit/s transfers it in 40 seconds when overhead and interruptions are already excluded. A nominal link rating alone would not establish that actual payload throughput.

In networking language, bandwidth commonly describes a data-transfer capacity, while throughput is the achieved rate and latency a delay. Radio-frequency bandwidth measured in hertz is a different use of the word, so units matter. A path can carry large files efficiently yet have a noticeable response delay. The internet links networks; the web is an information/application system that uses that infrastructure. Email and other services show why internet and web are not identical.

Cloud computing provides network access to shared configurable computing resources such as storage and processing. It still relies on physical servers, networks, electricity and cooling. An Internet-of-Things service connects sensing and communication to processing and sometimes actuation: a sensor measures a condition, information is interpreted, and a device or person may respond. A miscalibrated sensor, unreliable connection or unsuitable decision rule can undermine the whole service even if every component is described as smart.

ORIGINAL WORKED EXAMPLE

Convert bytes before dividing by a bit rate

A 120 MB file uses decimal MB=10^6 bytes and eight bits per byte. The sustained payload throughput is 24 Mbit/s, with overhead already excluded and no interruptions.

  1. Decimal 120 MB contains 120×10^6 bytes. Multiply by 8 bits/byte to obtain 960×10^6 bits, or 960 Mbit.
  2. At sustained payload throughput 24 Mbit/s, time=960/24=40 s. The common mega factor cancels.
  3. Overhead is already excluded and interruptions are absent by assumption. A nominal advertised link rate or propagation latency alone would not establish this result.

Answer 40 seconds.

ORIGINAL PRACTICE · 1

Does moving data to the cloud eliminate computers and electricity use?

Reveal answer & reasoning

No.

Remote data centres and networks provide the physical service.

05 / UNDERSTAND & APPLY

Identity, permissions and data protection

Authentication establishes confidence in a claimed identity or control of the relevant authenticator. Authorisation determines which resources or actions are permitted. A verified employee can legitimately be allowed to enter a system but denied payroll records because the role does not require them. Successful login therefore does not imply universal access. These are distinct questions even though practical systems combine them in one user experience.

Encryption protects confidentiality under its key and security assumptions. Integrity mechanisms help detect unauthorised alteration, access controls restrict actions, and backups support recovery after loss or damage. No single measure performs every function: encrypting a file does not make a false input true or restore a deleted file without a recoverable copy. A backup’s usefulness must be established by appropriate restoration checks, not merely by the existence of a backup label.

Privacy also concerns why data are collected, how much is necessary, how long they are retained and who is accountable. A technically secure database can still collect excessive personal information or use it for an unjustified purpose. A service should evaluate proportionate data needs and access, rather than retain everything because future analytics might find a use. These are defensive governance concepts, not attack instructions or claims about a specific current legal duty.

ORIGINAL PRACTICE · 1

A verified employee can sign in but is denied payroll access. Is that contradictory?

Reveal answer & reasoning

No.

Identity authentication and role-based authorisation answer different questions.

ORIGINAL PRACTICE · 2

Can encryption replace a tested backup after accidental deletion?

Reveal answer & reasoning

No.

Confidentiality and recoverability are distinct properties.

06 / UNDERSTAND & APPLY

Artificial intelligence and evidence limits

Machine learning fits patterns from data to support tasks such as prediction or classification. Good performance on training examples can fail to generalise when a model has learned incidental detail, when the sample excludes important cases or when deployment conditions change. Overfitting and distribution change are therefore different routes to unreliable new-case performance. Evaluation needs representative unseen cases and ongoing attention to the setting, not just a large training accuracy figure.

Accuracy counts correct classifications among all cases, while recall for a target class counts detected target cases among all actual target cases. In the fictional set of 1000 cases, 900 are ordinary and 100 need attention. A classifier calling every case ordinary is 90% accurate but detects none of the attention cases, giving 0% recall for that class. It predicts no positives, so positive predictive value is undefined because its denominator is zero. The application’s purpose determines which failures matter.

A numerical model score is not automatically a calibrated probability: calibration asks whether stated probabilities match observed frequencies under relevant conditions. Nor does prediction establish causation. Generative systems can produce plausible but unsupported statements, making independent verification important. Human accountability, appropriate evaluation and a way to challenge consequential errors remain necessary; an AI label does not establish correctness or justify handing every high-impact decision to a model.

ORIGINAL WORKED EXAMPLE

A misleadingly accurate classifier

A fictional classifier evaluates 1000 cases: 900 are ordinary and 100 need attention. It labels every case ordinary.

  1. Calling every case ordinary correctly labels the 900 ordinary cases out of 1000, so overall accuracy is 90%.
  2. None of the 100 attention cases is detected, so attention-class recall=0/100=0%. All 100 are false negatives for that target class.
  3. There are no predicted positives. Positive predictive value would be true positives divided by all predicted positives, giving a zero denominator and therefore no defined value. The metric cannot be reported as 100% simply because false positives are absent.

Answer 90% accuracy, 0% attention-class recall; positive predictive value undefined.

ORIGINAL PRACTICE · 1

Does perfect training accuracy guarantee equal performance on a new population?

Reveal answer & reasoning

No.

Overfitting, different measurement and distribution shift can invalidate that inference.

07 / UNDERSTAND & APPLY

Satellites, navigation and remote sensing

A launch vehicle transports a payload, while an orbiting satellite performs its mission after reaching the intended trajectory. Orbital motion combines tangential motion with gravitational acceleration; being in orbit does not mean gravity is absent. Orbit choice affects coverage, revisit time and communications delay. A lower orbit can reduce distance-related signal delay but does not remain over one ground location; a mission’s purpose determines the tradeoff.

Geostationary orbit is a special circular, equatorial, prograde orbit with the appropriate sidereal period, approximately 23 hours 56 minutes. It appears fixed relative to the rotating Earth in the ideal model. Geosynchronous means matching that rotation period but does not alone ensure the required inclination and circular geometry. Polar or near-polar paths instead pass over very different latitudes as Earth rotates, supporting other coverage patterns. No current mission status or exact satellite count is needed to explain the distinction.

Passive sensors measure naturally available reflected or emitted radiation; active sensors transmit a signal and measure a return. Reflected visible-light surface imagery requires illumination and can be obstructed by clouds. Suitable microwave radar supplies its own illumination and can operate at night and through many clouds, but wavelength, rain, geometry and surface properties affect the signal. Radar does not directly measure every variable of interest, and interpretation still needs calibration and validation.

Navigation receivers use timed signals and satellite-orbit information to estimate position. In the standard unconstrained three-dimensional solution, the unknowns include three position coordinates and receiver-clock offset, requiring at least four suitable satellite signals. This is a model condition, not a claim that every aided or constrained navigation system always needs exactly four. Signal blockage, geometry, propagation effects and timing errors affect accuracy. A satellite signal is not itself proof that a displayed location is exact.

ORIGINAL WORKED EXAMPLE

Select a sensor for a cloudy night

A flood team needs surface observations during a cloudy night. Compare reflected-visible passive imaging with suitable active radar, assuming both platforms cover the area.

  1. Reflected visible-light imaging requires an illuminating source; the supplied cloudy night lacks sunlight and clouds obstruct surface information.
  2. Suitable active radar provides its own illumination and can obtain returns through many clouds, so it is the stronger option under the given coverage assumptions.
  3. The return still depends on wavelength, geometry and surface properties. Interpretation and independent validation are needed; radar is not a universal direct measurement of all flood variables.

Answer Suitable radar is the stronger supplied option, with interpretation and validation limits.

ORIGINAL PRACTICE · 1

Is every geosynchronous satellite geostationary?

Reveal answer & reasoning

No.

Geostationary additionally requires the appropriate equatorial circular prograde geometry.

08 / UNDERSTAND & APPLY

Biotechnology and genetic information

Biotechnology uses biological processes or components to provide products or services. Fermentation can use microbial metabolism to transform substrates, as in some food production; diagnostics can detect biological signals; genetic engineering changes genetic material. These are related applications but not synonyms. A familiar fermentation process is not evidence that every biotechnology involves altering genes.

Genetic engineering deliberately alters genetic material. Genome editing can change an existing sequence without necessarily adding a gene from a different species. Sequencing reads the order of DNA bases and does not itself edit that sequence. A readout may support research or analysis, but its interpretation and any later intervention are separate steps. The lesson provides no laboratory protocol, pathogen manipulation or treatment recommendation.

Assess each application through its trait, organism, exposure route and environment. A useful engineered trait may still need evidence about ecological effects, unintended outcomes, access and oversight; a natural origin is not a guarantee of safety either. Consent and rights matter when human data or interventions are involved. Broad labels cannot replace application-specific evidence or an appropriate biosafety and governance process.

ORIGINAL PRACTICE · 1

Does sequencing DNA necessarily alter that DNA sequence?

Reveal answer & reasoning

No.

Reading information and editing it are different operations.

09 / UNDERSTAND & APPLY

Nanoscale properties and safety

Nano denotes one billionth: one nanometre is 10^-9 metre. Nanotechnology commonly concerns approximately 1–100 nm dimensions where size and structure can produce useful behaviour. A size label alone does not prove that an object has every special property associated with some nanomaterials. Composition and structure remain important.

For geometrically similar solid cubes, surface-area-to-volume ratio is 6/a, where a is the side length. Reducing a from 100 nm to 20 nm raises the ratio from 0.06 nm⁻¹ to 0.30 nm⁻¹, a fivefold increase. More accessible surface can change interaction with the surroundings; some size-dependent behaviour also involves quantum effects. Aggregation, coatings and chemistry can alter the accessible surface and performance. The geometric ratio does not by itself quantify toxicity or catalytic benefit.

Nanomaterial risk depends on the material, exposure route, dose, persistence and the transformations it undergoes. A useful filter or sensor still needs scrutiny of production, particle release during use, recovery and disposal. End-of-life management matters because materials can enter air, water, soil or waste streams after the intended service ends. Neither all nanoparticles are dangerous nor all nanoscale products are safe is a defensible universal claim.

ORIGINAL WORKED EXAMPLE

Smaller cubes and accessible surface

Compare solid cubes of the same material with side lengths 100 nm and 20 nm; ignore aggregation and surface coatings.

  1. A cube has surface 6 a² and volume a³, giving surface/volume=6/a. At 100 nm the ratio is 0.06 nm⁻¹; at 20 nm it is 0.30 nm⁻¹.
  2. Divide 0.30 by 0.06 to obtain a fivefold increase. Identical shape and material are held fixed in the supplied geometry.
  3. Accessible surface in a real material can be altered by aggregation or coatings. The calculation does not establish a fivefold increase in toxicity, useful reaction rate or any other outcome.

Answer Fivefold higher surface-area-to-volume ratio for the smaller cube.

ORIGINAL PRACTICE · 1

Does a tenfold decrease in geometrically similar cube side length give a tenfold surface-to-volume increase?

Reveal answer & reasoning

Yes under the stated geometry.

6/a increases inversely with a; practical aggregation can change accessible surface.

10 / UNDERSTAND & APPLY

Choosing evidence rather than a technology slogan

Compare technologies using the same service and lifecycle boundary. For a digital public service, operational computing and communications use energy, while servers and devices require materials, manufacture, replacement and end-of-life management. These demands should be weighed against benefits such as avoided travel, reduced waste or improved access, rather than assumed away. Repairability and a useful lifetime can change the result as much as a component’s rated efficiency.

In Prelims reasoning, test an absolute statement against mechanism and conditions. A fuel cell’s water output does not prove zero lifecycle emissions; a cloud service still uses infrastructure; a four-satellite navigation rule has an explicit model boundary. Avoid reversing every overclaim into an equally absolute dismissal of innovation. A sound answer identifies the supported benefit, the condition needed for it and the evidence still required. The accompanying matrix condenses four examples whose full explanations remain in the prose.

Full-size scientific diagram. Swipe or scroll horizontally; keyboard users can focus the diagram and use arrow keys.

Four technology mechanisms and their evaluation limitsOriginal table with semiconductor gate control, radar transmitted-return sensing, biotechnology biological/genetic methods and nanoscale size/structure interactions. Each row pairs its mechanism with necessary evidence or boundaries. All content repeated in prose; no quantitative scale or universal technology ranking.Mechanism → application → conditionFour conceptual examples; the full explanations and safeguards appear in the lesson.FieldMechanism and possible useCondition or limitationSemiconductorGate field controlschannel currentRequires circuit energyand device designRadar sensingTransmitted signaland measured returnWavelength, geometryand validation matterBiotechnologyBiological processor genetic changeTrait, exposure, consentand biosafetyNanomaterialSize and structurealter interactionsComposition, aggregationand lifecycleA useful mechanism does not by itself prove universal benefit, safety or reliability.
Identify the mechanism, then test the application under its actual conditions.

KEEP THE BIG IDEAS

Quick revision

  • A demonstration needs deployment evidence before a service claim follows.
  • Semiconductor control changes conduction without creating energy.
  • Energy capacity and power capability are different.
  • Authentication and authorisation answer distinct questions.
  • High overall accuracy can hide failure on the important class.
  • Sensing and navigation claims depend on geometry and signal conditions.
  • Sequencing reads DNA; editing changes it.
  • Nano-scale geometry does not alone determine safety or performance.
Back to the top ↑

READ WITH CONTEXT

References & scope

The teaching notes, passages, worked explanations and practice are original. Official references identify the assessment framework and skill scope; they do not endorse this lesson or license official question reproduction.

Explore this topic in context

UPSC analytics cover only explicitly reviewed printed papers. Paper-specific parent counts include unresolved classifications; subparts and Essay choices remain separate. Teaching availability is independent of historical observations.

KEEP EXPLORING

Your next idea is waiting.

Back to Civil Services General Studies index →Browse all topic notes