How artificial intelligence works
This briefing paper from the European Parliament's Panel for the Future of Science and Technology (STOA) provides an accessible introduction to how artificial intelligence works, covering the chronology of its development from symbolic AI to data-driven machine learning and speculative future developments. It examines core techniques such as expert systems, artificial neural networks, deep learning, back propagation, gradient descent, and evolutionary training methods. Additionally, the briefing discusses related concepts like big data, computer vision, natural language processing, generative adversarial networks, robotics, and quantum AI, helping readers engage in clear-headed reflection on AI opportunities and challenges.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
Members and staff of the European Parliament, as well as the general public interested in understanding the foundational mechanics of artificial intelligence.
What the document contains
The document explains the core technical concepts and historical evolution of artificial intelligence, covering symbolic AI, machine learning, neural networks, training methods, data practices, and speculative future technologies.
The text does not provide binding regulatory obligations, legal definitions for enforcement, or specific policy recommendations for governing AI systems.
- Artificial intelligence is an umbrella term covering various methods, ranging from early symbolic rules to modern data-driven machine learning.
- Symbolic AI uses precise human-crafted rules and keeps humans in the loop, making it transparent but limited in complex, dynamic environments.
- Data-driven machine learning relies on massive data availability and training algorithms like artificial neural networks to autonomously improve performance.
- Deep learning utilizes multi-layered neural networks to extract abstract features, though training requires careful balancing of design parameters and ethical data sourcing.
- Future speculative developments include artificial general intelligence, quantum AI, brain emulation, and advanced robotics.
· First wave: Symbolic artificial intelligence · Second wave: Data-driven machine learning · The art of artificial intelligence · Making sense of the world: Identifying language, images and sounds · Imagination and creativity: Producing language, images and sounds · Future waves: Towards artificial superintelligence?
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Accountability
0/5Clear responsibility for AI outcomes
N/A
screening score · 2026-10-23
Fairness
0/5Non-discrimination and equitable treatment
N/A
screening score · 2026-10-23
Human Dignity
2/5Respect for the inherent worth of persons
Would it have consciousness , feel pain or sadness ? Would deleting it be murder? Would owning it be slavery? Would modifying it be abuse?
Brain emulation and artificial consciousness
screening score · 2026-10-23
Human Oversight
3/5Humans remain in control of consequential decisions
Symbolic AI can be said to ' keep the human in the loop ' because the decision- making process is closely aligned to how human experts make decisions.
First wave: Symbolic artificial intelligence
screening score · 2026-10-23
Privacy
1/5Protection of personal data and private life
Other data might be subject to copyright, or require ' informed consent ' from the owner before it is used for research or other purposes.
The art of artificial intelligence
screening score · 2026-10-23
Transparency
3/5Openness about how AI systems work and are used
Furthermore, humans can easily understand how these systems make specific decisions. They can easily identify mistakes or find opportunities to improve the programme
First wave: Symbolic artificial intelligence
screening score · 2026-10-23
Enforcement
0/5Consequences and compliance mechanisms with real force
N/A
screening score · 2026-10-23
Impact Assessment
0/5Required assessment of an AI system's risks/impacts before or during use
N/A
screening score · 2026-10-23