Ethics and ownership

What ethics means for a computing professional

Ethics are the principles a person uses to decide what is right and wrong in the work they do. Professional ethics are the same idea applied to a job: the standards of behaviour expected of someone who has specialist knowledge, and whose decisions affect people who do not share that knowledge. A programmer knows what the code does; the customer, the patient and the member of the public do not, and they have to trust the programmer to act in their interest.

Computing needs its own ethical standards because of the power a programmer holds and the scale at which that power operates. One line of code in a library used by thousands of programs can affect millions of people, and one badly designed system can leak the personal data of a whole population. The harm is often invisible and delayed: a copy of code produces no immediate symptom, and the person who suffers is rarely in the room when the decision is made. Law also lags behind: legislation is written after a technology exists, so there are many decisions that are perfectly legal but still wrong, such as quietly selling a customer's data because the terms and conditions allowed it.

So the purpose of ethics for a computing professional is to give a standard of behaviour that applies even where no law exists, to protect the public from harm that they cannot detect for themselves, to keep trust in systems that people have no choice but to rely on, and to give a professional the confidence and the language to refuse an instruction that would harm someone. Ethics is not the same thing as personal opinion, and it is not the same thing as legality: a good exam answer says what the professional should do, why the law does not settle it, and who is affected.

A useful way to hold the whole topic together is that ethics, professional bodies, copyright and licensing are all about the same question at different scales - what is the right thing to do with somebody else's work and somebody else's data. Copyright and licences settle some of it in law, a code of conduct settles some of it in professional expectations, and ethics covers what is left.

• Ethics: principles for deciding right from wrong conduct in a professional role.
• Purpose: to protect the public where the law is silent or slow, because users cannot see what the code does.
• Why computing needs it: a program acts at scale, harm is often invisible, the professional knows more than the client, and technology changes faster than legislation.
• Ethics is not legality: an action can be legal and still unethical, and a professional is expected to meet the higher standard.
• The four groups usually affected: the public, the client or customer, the employer, and the profession itself.

Professional bodies: the BCS and the IEEE

A professional body is an organisation that represents, regulates and supports the people who work in a profession. Computing has two that the syllabus names. The BCS (British Computer Society), which calls itself The Chartered Institute for IT, is the UK professional body for computing; it sets a Code of Conduct for its members, awards the title Chartered IT Professional (CITP), publishes magazines and research, runs training and events, and can discipline a member who breaches its code. The IEEE (Institute of Electrical and Electronic Engineers) is a much larger, US-based professional organisation for engineers, including computer engineers and computer scientists; it publishes technical standards and journals, runs conferences, and has its own Code of Ethics, which asks members to accept responsibility for decisions that affect the safety, health and welfare of the public.

Membership is voluntary, and that is the point: nobody is forced to join, so joining is a public statement that you accept a published standard of behaviour and are willing to be held to it. In return the member gets credibility with employers and clients, because membership is evidence of competence and of a commitment to keep it up to date; continuing professional development, because both bodies provide training, publications and events; guidance when a situation is difficult, because a code gives a member something to point to when an employer asks for something wrong; and a community of people doing the same work, which is where technical standards and best practice come from.

The other side of membership is accountability. Both bodies publish a code and both can investigate complaints and suspend or expel a member who breaks it. That sanction is what turns a code from a suggestion into something real: a member who ignores it can lose the professional title that their career depends on, and the employer and the public can see that the profession takes the standard seriously. This is also why the bodies matter to people who never join them - a public code raises the expected standard for everybody in the industry.

Exam answers should name a body, say what its code requires in the situation described, and say what membership adds: it makes the standard enforceable and visible, it gives the member support and a reason to keep their skills current, and it protects the public by giving them a route to complain.

Acting ethically, and the impact of acting unethically

Most exam questions here give you a situation and ask you to say what the professional should do and what difference it makes. Work through it in a fixed order: who is affected (the user, the public, the client, the employer, colleagues, the profession), what the rules say (the law, the contract, the code of conduct), what the realistic options are, and what the consequences of each would be. Then decide, and justify the decision by the harm avoided rather than by how the professional feels.

Here is the standard scenario. A developer tests her program and finds that in one set of circumstances it will corrupt the customer's stored data, and her manager tells her to release it anyway because the customer has a deadline and the fault is unlikely. Acting ethically means reporting the fault in writing, refusing to sign the release off and escalating it to a more senior manager or to the customer: the cost is a slipped schedule and an awkward conversation, but the users' data is protected and the company is not exposed to compensation claims, lost contracts and reputational damage. Acting unethically means releasing quietly and hoping: the deadline is met, but data is lost, the fault costs far more to fix once the system is live, users may be harmed, and both the developer and the company can be held responsible.

The same structure works for data. If a company proposes to sell its customers' personal data to advertisers because the terms and conditions technically permit it, the ethical professional points out that consent was not informed, that customers would not expect it, and that the law on data protection may still restrict it; the unethical professional implements it because it is profitable and nobody has objected yet. Notice that in both cases the harm falls on people outside the room, which is exactly why the professional standard has to replace the customer's ability to see what is happening.

A strong answer also acknowledges the cost of acting ethically. It can mean a delay, a lost contract, an awkward relationship with a manager, or, in the worst case, losing a job. Say so, then say why the professional does it anyway: the code of conduct, the duty to the public, and the fact that the long-term damage of the unethical option is larger than the short-term cost of the ethical one.

Why copyright legislation is needed

Copyright is a legal right that gives the creator of a work control over how that work is copied, adapted, published and distributed. It protects the expression of an idea - the actual words, the actual lines of code, the actual image or recording - and not the idea itself, which is why two programmers can each write a program that sorts a list without infringing each other. In the UK, a computer program is protected as a literary work, and protection lasts for the author's life plus 70 years; you do not have to register anything, the right exists as soon as the work is recorded in a fixed form.

Legislation is needed because without it there is nothing to stop anyone copying the work of another person and selling it. A program can take years and a great deal of money to write, and it can be copied in a second at no cost and with no loss of quality, so if copying were free and lawful the person who paid for the work would always be undercut by the person who copied it, and there would be no reason to invest in writing software, music, books or films. Copyright exists to reward the creator, to make that investment worth making, and so to keep new work being produced - an economic argument as much as a moral one.

The law also gives the creator control: the right to decide who may copy the work, to charge for that permission, to require that they are credited, and to object if the work is altered and presented as something else. That control is what a licence uses. The copyright owner grants permission on stated terms, and the licence is the document that records the terms - how many machines may run the software, whether the source code is supplied, whether it may be modified, whether modified copies may be passed on, and whether it may be used commercially.

Two points are examined again and again. Buying a copy of software does not buy the copyright; you own the disc or the download and a licence to use it, and copying it to sell to others is infringement, commonly called software piracy. And open source software is still copyrighted - the copyright is simply licensed on generous terms, so the freedoms the user has come from the licence, not from the absence of copyright. Free of charge and free of restriction are different things.

• Protected: the expression - source code, object code, text, images, music, film, databases.
• Not protected: the idea, the algorithm, the function or the look of a program's behaviour.
• Why needed: copying is instant, free and perfect, so without the law nobody would pay to create the work; copyright rewards the creator and protects the investment.
• What it gives the owner: the right to control copying, adaptation, distribution and commercial use, and to be paid for granting permission.
• What it does not give the buyer: the copyright itself. Buying software buys a licence to use it, not ownership of the work.
• Infringement: copying, distributing or adapting the work without permission, including installing one licensed copy on many machines or sharing it online.

The four types of software licence

A software licence is the legal agreement under which the copyright owner allows someone to use the software, and it sets out exactly what that person may and may not do. Every licence on the syllabus is a different answer to the same questions: may I run it, may I see and change the source code, may I give copies to other people, and may I sell what I make with it?

Free Software Foundation (FSF). The FSF, founded by Richard Stallman, defines free software by four freedoms: the freedom to run the program for any purpose, the freedom to study how it works and change it (which requires the source code), the freedom to redistribute copies, and the freedom to distribute modified versions so the community benefits. Free means freedom, not price; software can be free software and still be sold. Its best-known licence is the GNU General Public License (GPL), which is copyleft: anyone who distributes a modified version must license it under the GPL too and make the source available. That is deliberate - it stops someone taking community code, improving it and locking the improvement away in a closed product. The Linux kernel is the largest example of GPL software.

Open Source Initiative (OSI). The OSI was founded in 1998 and maintains the Open Source Definition, a set of criteria a licence must meet (free redistribution, source code available, allow modifications and derived works, no discrimination against people or fields of use). It approves licences rather than writing one. Some approved licences, such as the GPL, are copyleft; others, such as MIT and Apache 2.0, are permissive, which means they allow the code to be used in a closed-source commercial product as long as the copyright notice is kept. So the FSF and the OSI overlap but are not the same: the FSF argues from the user's freedom to control their own computing, the OSI argues from the practical and business benefits of open development. A licence can satisfy both.

Shareware is software distributed free of charge so that people can try it before they buy it. The copy is usually a full working program for a limited period, or a version with some features disabled, and after the trial the user is expected to pay a fee to keep using it or to unlock the rest. The copyright stays with the author, who usually asks that copies are passed on unaltered, so shareware is neither open source nor public domain; it is a marketing method for selling a proprietary product. Commercial (proprietary) software is sold under a licence - usually an end-user licence agreement (EULA) - for a price, per user, per device or by subscription. The source code is not supplied, modification and redistribution are prohibited, and in exchange the customer usually gets support, updates and a legal assurance that the software works as described.

Justifying a licence for a given situation

A justification question is not asking which licence is best. It is asking which licence fits what this user needs to do, and the answer is scored on the terms of the licence, not on price. Read the scenario for four things: does the user need to change the code, does the user need to give copies away or sell the result, does the user need to keep their own code secret, and does the user need support and a contract?

If the answer to the first two is yes and the user is happy to share their changes, free software under the GPL fits: it is legal to copy it onto as many machines as they like, they can adapt it, and they must publish their improvements. If the user wants to modify the code and put it inside a product whose own source stays secret, the GPL does not fit, because copyleft would force the whole derived work open; a permissive open source licence such as MIT or Apache 2.0, or a commercial licence for a proprietary library, does. If the user simply wants other people to try the program before paying, shareware is the distribution model. If the user is a business that needs guaranteed support, updates and legal protection, a commercial licence with a support contract is the right choice even where a free alternative exists, because the cost of the licence buys certainty.

There is a fifth consideration that separates a good answer from a full-mark one: the consequences of the licence for the user's own work. A charity that builds on GPL code must release its own additions; a company that embeds GPL code in a product may be forced to publish its source or to rewrite; a company that embeds MIT-licensed code can keep its own code closed as long as it keeps the notice. Say which of these the scenario can live with.

Finally, remember that a licence can be mixed: a project may use an open source library, pay for a commercial database and distribute its own front end under a shareware-style trial. The exam question gives one situation, so choose the licence that matches the situation's main requirement and justify it by referring to the terms you have just been given.

• Step 1: does the user need the source code and the right to modify it? If not, a commercial licence is simplest.
• Step 2: must the user be able to copy it freely or give it to others? If yes, an open licence (GPL or permissive) rather than a per-seat commercial one.
• Step 3: must the user keep their own code closed? If yes, avoid copyleft (GPL); use a permissive licence or a commercial one.
• Step 4: is support, an update path and legal liability part of what is being bought? If yes, a commercial licence is justified even when free software exists.
• Step 5: state the fit in one sentence - 'the GPL is justified because the school may install it on any number of machines and may adapt it, on condition that any version it passes on is released under the same licence'.

Artificial intelligence: what it is and what it is used for

Artificial intelligence (AI) is the name for computer systems that carry out tasks which normally need human intelligence: learning from examples, recognising patterns, understanding language, making judgements and taking actions in the physical world. The important difference from ordinary software is that a conventional program follows rules a programmer wrote, while an AI system is trained on data and works out its own rules from the patterns in that data. That is why the same system can recognise faces or translate languages without anybody writing a rule for every case, and also why it can be wrong in ways nobody predicted.

The main branches are worth knowing by name. Machine learning trains a model on large amounts of data so that it can classify or predict, and it underlies spam filters, credit scoring, product recommendations and medical image analysis, while expert systems store the knowledge of specialists as rules in a knowledge base with an inference engine and are used for diagnosis and for advice on complex procedures. Natural language processing handles human language in speech recognition, machine translation, chatbots and voice assistants, computer vision interprets images and video for facial recognition, number-plate reading, inspection and medical scans, robotics combines sensing, planning and movement in industrial robots, drones and driver-assistance systems, and generative AI produces new text, images, code or audio from patterns learned from existing material.

The applications follow from those branches: medical diagnosis and drug discovery, fraud detection in banking, route planning and autonomous transport, machine translation, personalised recommendations, agricultural monitoring from satellite and drone images, predictive maintenance on machinery, and chatbots handling customer service. One well-known example is AlphaFold, developed by DeepMind, which predicts the three-dimensional structure of a protein from its amino acid sequence - a task that used to take years of laboratory work per protein and that matters for understanding disease and designing medicines.

Keep the definition tight in an exam: AI is software that learns or reasons from data to perform tasks that need human-like intelligence, and it is trained rather than explicitly programmed. A program that follows fixed rules is not AI just because it is complicated.

• Machine learning: learns patterns from data instead of following rules written by a programmer.
• Expert systems: knowledge base plus inference engine, used for diagnosis and advice.
• Natural language processing: speech recognition, translation, chatbots, voice assistants.
• Computer vision: facial recognition, number-plate reading, inspection, medical imaging.
• Robotics: industrial robots, drones, driver assistance and autonomous vehicles.
• Generative AI: produces text, images, code or audio from patterns in training data.
• Applications: diagnosis, fraud detection, translation, recommendations, route planning, predictive maintenance, customer service.

The impact of AI: social, economic and environmental

Impact questions are marked by category and by who is affected, so sort your points three ways and say which is which. Social impacts concern people and how they live: access to services, privacy, fairness, safety, employment conditions and the quality of decisions made about them. Economic impacts concern money, work and markets: productivity, jobs, the cost of goods and services, competition and who captures the value. Environmental impacts concern energy, resources and waste.

Social. On the positive side, AI gives round-the-clock services that do not need a queue, translates languages well enough to open information to people who could not read it before, reads medical scans and flags disease earlier than a tired radiologist might, and provides voice interfaces and image descriptions that make technology usable for people with disabilities. On the negative side, an AI system is only as fair as its training data: if the data under-represents a group, the system is less accurate for that group, and that has happened in facial recognition and in recruitment screening. AI makes surveillance cheap and continuous, and it makes convincing fake images, audio and video easy to produce, which damages trust in evidence. Automated decisions about loans, benefits or policing are hard to challenge because the reasoning is opaque, and responsibility is easily lost between the developer, the data supplier and the organisation using the system.

Economic. AI raises productivity: tasks that took hours are done in seconds, which lowers costs and can lower prices. It creates new industries and jobs - data work, model development, AI auditing and the maintenance of the systems - and it lets small firms do work that used to need a large team. Against that, it displaces work, and not only manual work: routine cognitive tasks such as basic translation, first-line customer service, routine report writing and some coding are exactly what these systems do well, and the people who lose those jobs are often those least able to retrain. The cost of the computing power and data needed to train the largest models concentrates the market in a few very large firms, which raises prices and reduces choice. And the gains are unevenly shared: a company that replaces staff with AI captures the saving, while the cost of retraining and of supporting the displaced worker falls on the public.

Environmental. AI can reduce environmental damage: models forecast weather and flood risk, optimise the heating and cooling of buildings, plan delivery routes to cut fuel, and support precision farming that applies less fertiliser and water. But training and running large models uses very large amounts of electricity, and data centres need water for cooling and land for the buildings; where that electricity comes from fossil fuels, the emissions are real and growing. The hardware is replaced every few years, which produces electronic waste, and the chips and batteries it depends on require mining of scarce minerals, with the water use, land damage and pollution that mining brings. The balance depends on where the electricity comes from and what the AI is used for - a model that cuts the energy use of a national grid may repay its own cost, while one generating novelty images may not.

The conclusion the syllabus wants is not 'AI is good' or 'AI is bad'. It is that the impacts are unevenly distributed and depend on who controls the system, what data it was trained on, and whether anyone is accountable - which takes you straight back to professional ethics, codes of conduct and the argument for regulation such as the EU's risk-based AI Act.