Chapter 2 lays the technological foundations for the analysis developed in the rest of the book. In contrast to Chapter 1, which is largely AI-free, this chapter is deliberately entrepreneurship-light, focusing instead on the technological and socio-economic features of artificial intelligence. After a brief introductory overview of AI, the chapter provides a structured and accessible representation of deep learning technologies, explaining how they operate, what their core capabilities are, and where their main limitations lie. The chapter presents AI as a digital ‘statistical machine’ and examines the inferential architectures underlying contemporary deep learning systems, with particular attention to data-driven learning effects and feedback loops. Emphasis is placed on the socio-economic implications of these technologies for the functioning of markets, including issues related to data, infrastructures, and computational resources. Building on this foundation, the chapter then introduces the functional capabilities enabled by AI – predictive, generative, and agentic – and discusses how they are deployed within the economic system, illustrating the structure of AI supply chains. By isolating the technological dimension from entrepreneurial dynamics, the chapter provides a clear analytical baseline for understanding how AI reshapes economic environments before turning explicitly to entrepreneurship in subsequent chapters.

Chapter 5 analyses how to design more trustworthy and safer Clever Biometric Computing systems, that is, sensor technologies that read and computationally interpret the human body using AI. The complex socio-technical histories around facial recognition technologies (FRT) and emotion recognition (ERT) are presented, as key examples of Clever Biometric Computing. Discussion focuses on why they are ethically, legally, and socio-technically problematic. The chapter then explores the EU policy landscape around trustworthy AI and the EU Artificial Intelligence Act (AIA). The AIA formal legal requirements for design of AI systems are comprehensively analysed, particularly the turn to design for regulating high-risk AI systems (HRAIS). Some requirements that are analysed include human oversight, fundamental rights impact assessments, data governance, record keeping, and transparency. Different examples are presented too, such as ERT for safety in cars, industrial workplaces, and law enforcement use of live FRT. Ways to overcome challenges of operationalising the HRAIS requirements from an Accountable Design perspective are a key part of this chapter.

Chapter 1 provides an overall introduction to the book and its underlying motivations. It proposes and defines the concept of Clever Computing. It presents the historical emergence of the Internet of Things, AI, and Ubiquitous Computing, and analyses the risks of technological visions of the future which often neglect current social and human needs in favour of techno-led solutions. It explores the need to account for human needs in design and, by extension, legal values. The term Clever Computing is contextualised by exploring different examples of how AI and IoT are increasingly converging. This focuses on the ways AI is changing and improving interaction techniques with IoT; how AI is being used to make sense of IoT data and enabling distributed data analysis; and how AI is augmenting the existing functionality and increasing the autonomy of IoT systems.

This chapter brings together three practitioners who approach human-centered ML from creative, design, and research perspectives. Artist Memo Akten reflects on a decade of building generative systems that preserve what he terms "meaningful human control" and on how creative AI practice is shifting from generation toward curation. Designer Josh Lovejoy, who co-founded People+AI Research (PAIR) at Google and Microsoft’s Ethics and Society team, frames accountability as a design property: the question is not what the system needs in order to be accountable, but what users need in order to take accountability themselves. He also examines the misaligned incentives driving current AI development, and the costs of designing for seamlessness at the expense of user agency. HCI researcher Wendy Mackay argues for replacing the "human in the loop" paradigm with what she calls the "computer in the loop," placing the algorithm in service of the user rather than treating users as inputs to better algorithms. Together, these conversations span artistic practice, design methodology, research funding, and the long-term social consequences of ML and AI.

AI can help address the access to justice crisis by assisting unrepresented litigants who cannot afford legal representation. Millions worldwide struggle to navigate complex legal systems designed for legal professionals, resulting in fundamental inequalities in access to justice. The chapter introduces how legal AI assistants can help unrepresented litigants understand their rights, complete legal documents, and navigate court procedures.

The analysis covers a range of technologies, including generative AI, machine learning, computational logic, document automation systems, and more.

Beyond technological tools, the chapter addresses critical barriers, including unauthorized practice of law restrictions, and emphasizes the need for institutional reforms. The chapter argues that legal AI assistants should be treated as public goods, similar to courts and schools, and that a coordinated effort between developers, legal practice bodies, courts, and policymakers is necessary to effectively democratize access to justice and ensure equality under the law.

Generative AI systems offer unprecedented capabilities for addressing access to justice challenges through their ability to understand and generate human-like text. This chapter examines how large language models (“LLMs”) can bridge the language gap between laypeople’s factual descriptions and legal concepts, providing personalized legal information and assistance with legal tasks.

The analysis explores several applications, including the automated provision of legal information using retrieval-augmented generation (“RAG”), streamlined legal intake processes, intelligent form completion assistance, and AI-enhanced mediation. These systems show promise for identifying legal issues from plain language descriptions, assisting with the drafting of legal documents, and supporting dispute resolution processes.

However, significant challenges remain, particularly regarding hallucinations where models generate convincing but incorrect information, algorithmic bias that may perpetuate existing inequalities, and transparency concerns inherent in black-box AI systems. While generative AI holds promise for democratizing legal assistance, implementation with robust safeguards, continuous evaluation, and human oversight is essential to ensure these tools enhance rather than undermine access to justice.

The language gap is a fundamental barrier preventing unrepresented litigants from understanding their legal rights and defenses. The chapter identifies three components: open texture, which reflects the inherent ambiguity and vagueness of natural language; legalese, characterized by its specialized vocabulary and structural peculiarities; and the abstraction of legal predicates, which require translation between everyday facts and legal concepts.

The analysis demonstrates how legal language functions as a distinct system where common words acquire uncommon meanings, precision conflicts with clarity, and flexible terms create interpretive challenges.

Recent advances in large language models offer promising tools for bridging this gap by providing translation capabilities between legal language and ordinary discourse; however, significant challenges remain for certain implementations regarding accuracy, hallucination risks, and unauthorized practice of law restrictions.

Online Dispute Resolution (“ODR”) emerged from the convergence of Alternative Dispute Resolution practices and digital platforms, addressing the need for more accessible dispute resolution mechanisms. ODR platforms reduce logistical burdens by eliminating travel requirements and enabling asynchronous communication, while helping to level the playing field between represented and unrepresented parties.

This chapter examines how AI enhances ODR through automated settlement calculations using game theory, negotiation analysis through discourse monitoring, and sentiment analysis to detect contentious exchanges. AI systems can also assist facilitators by analyzing negotiation patterns and suggesting when interventions may be needed.

Despite significant potential benefits, ODR faces challenges, including the digital divide, algorithmic bias, and jurisdictional enforceability. Well-implemented ODR platforms, supported by transparent standards and ethical AI principles, may significantly contribute to making dispute resolution more equitable and responsive to modern societal needs, while requiring careful attention to user-centered design and robust data protection measures.

As more people navigate justice systems independently, legal decision support tools for unrepresented litigants are becoming increasingly necessary. However, tools for unrepresented litigants present unique challenges compared with those designed for attorneys, and a significant barrier to access is the gap between legal and ordinary language.

This chapter examines the evolution of dialogue systems from early programs like ELIZA to modern conversational AI platforms and explores how legal chatbots can assist unrepresented litigants with court procedures and accessing legal information. The analysis addresses critical design considerations, including the unavoidable influence of digital nudging on user decision-making and the importance of user-centered design principles. Generative AI technologies show promise for improving case elicitation and bridging the language gap despite current limitations, such as hallucination risks.

No single AI system or other technology can address all the challenges unrepresented litigants face when navigating the legal system. This chapter highlights the strengths and weaknesses of integrating various technological approaches, including generative AI, computational logic, machine learning, document automation, and dialogue systems.

Through the hypothetical case of Jaime, a tenant facing unsafe and unlivable housing conditions, the chapter illustrates how integrated AI systems can guide unrepresented litigants through every stage of legal proceedings, from initial problem recognition to case resolution. The analysis shows how chatbots may help engage legal consciousness, generative AI may assist in bridging the language gap, computational logic can ensure procedural accuracy, and machine learning can provide case analysis and outcome prediction.

However, the chapter also addresses significant limitations of integrated approaches, including compounding errors across systems, cumulative biases, and substantial resource requirements for development and maintenance. While integrated AI systems offer promising potential for expanding access to justice, their practical implementation faces considerable financial and organizational constraints that may limit widespread adoption.

This chapter provides a foundational introduction to generative AI and large language models, addressing their capabilities and limitations. These models can perform competitively with humans on standardized exams and legal reasoning tasks and can be applied flexibly to legal challenges through natural language prompts.

However, significant limitations constrain their deployment in legal contexts. Models produce hallucinations, generating confident but incorrect information that can mislead users making critical legal decisions. The ELIZA effect compounds this risk as users tend to anthropomorphize AI and overestimate its capabilities. Additional challenges include potential biases introduced by the training data, privacy concerns associated with hosted models, and substantial computational costs. Understanding both the opportunities and limitations is crucial for safely deploying generative AI to address access-to-justice challenges.