Consequences of Computing

The growth of computing has fundamentally changed how people communicate, access information, and interact with institutions. This creates both enormous opportunities and serious risks.

Three areas of change the spec identifies:

Area	How computing changed it
Monitoring behaviour	Surveillance cameras, smart devices, app usage tracking, GPS, cookies — individuals' movements and actions can be recorded at scale
Amassing and analysing personal data	Social media platforms, governments, and corporations collect vast personal datasets; machine learning analyses patterns in this data
Distributing and disseminating information	Anyone can publish globally; misinformation spreads as easily as accurate information

These changes are not inherently good or bad — they enable medical breakthroughs, disaster response, and democratic participation, while also enabling surveillance states, discrimination, and manipulation.

Computer scientists and software engineers hold unusual power: code they write affects the lives of millions. With that reach comes responsibility.

Key principle: algorithms and software are not neutral. Every design decision encodes values:

• A loan-approval algorithm that uses historical data may perpetuate historical discrimination
• A content recommendation algorithm optimised for engagement may amplify outrage over accuracy
• A facial recognition system trained on limited demographic data may be less accurate for some groups

Responsibilities include:

• Being aware of the values and assumptions embedded in systems they build
• Considering potential harms before deployment, not only after
• Advocating for ethical design decisions even under commercial pressure
• Being transparent about what systems do and do not do

The scale of computing makes these responsibilities acute: a well-intentioned mistake in a widely deployed system can cause harm to hundreds of millions of people simultaneously.

Computing creates genuine opportunities alongside real risks:

Opportunities:

• Medical research: genomic databases, drug discovery, AI-assisted diagnosis
• Environmental monitoring: satellite data analysis, climate modelling, smart energy grids
• Accessibility: screen readers, real-time transcription, assistive devices
• Education: online learning, personalised tutoring, knowledge access in remote areas
• Democratic participation: e-petitions, open government data, civic platforms

Risks:

• Privacy: personal data collected without informed consent; data breaches
• Discrimination: biased algorithms in hiring, lending, criminal justice
• Misinformation: social media algorithms amplifying false content
• Job displacement: automation replacing routine cognitive and manual work
• Security: cyberattacks on critical infrastructure (power grids, hospitals)
• Surveillance: governments and corporations monitoring individuals at scale

The same technology that enables a GPS navigation app also enables tracking of individual movements. Context and intent determine whether a given use is beneficial or harmful.

Legislators face fundamental difficulties in regulating digital systems:

Why legislation struggles to keep pace:

• Technology evolves faster than the legislative process
• Software and data cross national borders instantly — jurisdiction is unclear
• Technical complexity makes it difficult for non-technical legislators to understand what they are regulating
• Large technology companies have significant lobbying influence

Specific challenges:

Challenge	Example
Jurisdiction	A social media platform headquartered in one country hosts content illegal in another
Anonymity	End-to-end encryption makes lawful interception technically difficult
Data ownership	Who owns personal data a company holds? The user? The company?
Algorithmic accountability	How do you audit a black-box AI decision for fairness?
Pace of change	GDPR was years in development; new AI capabilities emerged during its drafting

Key legislation examples (context only): the UK's Computer Misuse Act criminalises unauthorised system access; GDPR (UK and EU) regulates personal data handling; the Online Safety Act addresses harmful online content.

Algorithms do not operate in a cultural vacuum. They reflect the values and priorities of the people who designed them and the data they were trained on.

How values enter algorithms:

• Objective function: what the algorithm optimises for (clicks, accuracy, profit, fairness)
• Training data: historical data reflects historical inequalities — a hiring algorithm trained on past hires may replicate past biases
• Feature selection: what inputs the algorithm considers (postcode as a proxy for race)
• Threshold decisions: where to draw the line (false positive rate vs false negative rate in medical screening)

Cultural implications:

• A voice assistant designed and tested primarily by speakers of one accent will perform worse for others
• Translation systems trained on biased corpora propagate those biases
• Recommendation algorithms operating across cultures may impose one culture's norms on another

Computer scientists have an obligation to test their systems across diverse populations and acknowledge limitations honestly.

1. Giving only positive or only negative consequences

AQA questions on the consequences of computing expect balanced analysis. Technologies like social media, AI, and mass data collection have both beneficial and harmful consequences. A one-sided answer misses marks.

2. Stating "algorithms are neutral" or "technology is just a tool"

The spec explicitly states that algorithms and software embed moral and cultural values. This is a required concept: design choices (objective functions, training data, thresholds) are value-laden and have real-world effects.

3. Confusing legal and ethical issues

An action can be legal but unethical (collecting data users haven't meaningfully consented to) or illegal but ethically debated (sharing information about government wrongdoing). Questions asking about "ethical" issues are not satisfied by listing laws — they require discussion of right and wrong, harm, fairness, and responsibility.

4. Ignoring the scale argument

A key concept in this topic is scale: computing systems operate at a scale (billions of users) that makes design choices have societal-level effects. This is why the spec emphasises that CS professionals carry unusual responsibility — a bias in a physical process affects a few; a bias in a widely deployed algorithm affects millions.