Ethical and legal issues Revision Notes

Accountability and Legal Liability in AI

Accountability in AI refers to identifying which human or entity is responsible for an AI system's decisions and outcomes. This becomes complex when AI operates autonomously, making it difficult to pinpoint blame for errors or harms. Legal liability is the legal responsibility for the AI's actions, and current laws may not adequately cover AI's unique characteristics. For instance, if an autonomous vehicle causes an accident, is the manufacturer, software developer, owner, or even the AI itself liable? New legal frameworks are being explored to address this.

Safety and Algorithmic Bias

AI safety ensures that systems function as intended without posing risks to humans or infrastructure. This includes preventing unintended behaviours, vulnerabilities to attack, and system failures. Algorithmic bias arises when AI models are trained on skewed or unrepresentative data, leading them to make unfair or discriminatory predictions or decisions. This can manifest in areas like hiring, loan applications, or even criminal justice, where historical biases in data are unintentionally encoded into the AI's logic, perpetuating inequality.

Ethical Frameworks for AI Development

Developing ethical AI requires frameworks that guide its design, deployment, and usage. These frameworks often advocate for principles such as transparency (understanding how AI makes decisions), fairness (avoiding bias and discrimination), robustness (reliability and resilience), and privacy (protecting personal data). Adhering to these principles helps mitigate the risks associated with AI, machine learning, and robotics, promoting their beneficial use while safeguarding societal values.

Privacy and Ownership of Personal Data

Privacy is the right to control how personal information is collected, used, and shared. Ethical dilemmas arise when this right is compromised. Data ownership is legally complex; while individuals generally retain rights over their personal data, organisations often acquire processing rights through terms of service. Understanding these distinctions is crucial for responsible data handling.

Consent and Data Misuse

Consent is a fundamental legal requirement for processing personal data, meaning individuals must explicitly agree to the collection and use of their information for specified purposes. Misuse occurs when data is used beyond the scope of initial consent, or for purposes that are harmful or violate an individual's rights. This includes unauthorised access, sharing, or processing.

Data Protection Laws (e.g., GDPR)

Data protection refers to the legal and regulatory frameworks governing the handling of personal data. Laws like the General Data Protection Regulation (GDPR) in Europe mandate strict rules for how data is collected, stored, processed, and secured. These laws define individual rights, such as the right to access, rectify, or erase personal data, and impose obligations on organisations.

Copyright in Software

Copyright automatically protects original expressions of ideas, not the ideas themselves. For software, this typically covers the source code, object code, structure, sequence, and organisation of the program. It prevents unauthorised reproduction, adaptation, distribution, and public display of the software. Copyright arises automatically upon creation, without formal registration in many countries, though registration can strengthen enforcement options.

Patents for Computer Systems and Software

Patents protect inventions that are new, non-obvious, and useful. In the context of computer systems and software, this usually applies to novel algorithms, methods, or processes implemented in software that solve a technical problem in an inventive way. Obtaining a patent is a complex and lengthy process, often requiring significant investment. Patents grant exclusive rights to the inventor for a limited period, typically 20 years, to prevent others from making, using, or selling the invention.

Trademarks for Software Products

Trademarks protect branding elements like names, logos, and slogans that identify the source of goods or services. For software, a trademark might protect the name of an operating system, an application, or a software company's logo. Trademarks help consumers distinguish legitimate products from counterfeits and build brand recognition and goodwill. Registration is not always mandatory but provides stronger legal protection and rights.

Software Licensing

Licensing is a contractual agreement where the intellectual property owner grants permission for another party to use their software under specified conditions. This defines how the software can be used, distributed, or modified. Common types include end-user license agreements (EULAs), open-source licenses (e.g., GPL, MIT), and commercial licenses. Licensing is fundamental to software distribution, controlling access, and revenue generation while maintaining ownership.