98 Types of AI Models and Their Social Worlds

Introduction: Why AI Models Matter

Artificial Intelligence (AI) is often discussed as if it were a single unified technology. In reality, AI consists of many different types of models built using different technical approaches datasets and goals. In “Types of AI Models: A Deep Dive with Examples,” Kezia Jungco outlines six major categories of AI models: rule-based systems, machine learning models, deep learning models, natural language processing (NLP) models, computer vision models, and generative AI model. Each of the represents not only a technical distinction but also a different way of structuring knowledge, automating decisions, and distributing power.

From a science, technology, and society (STS) perspective, Ai models are not merely computational tools. They are technical systems embedded in social and economic structures, including cultural norms, institutional priorities, and historical inequalities. This chapter explores each of Jungco’s six AI model types while examining how they shape and are shaped by social forces.

AI Through an STS Lens

One useful framework for understanding AI in STS is social constructivism, as described by Andre Kukla in Social Constructivism and the Philosophy of Science. Social constructivism argues that scientific knowledge and technological systems are shaped by social processes, institutions, and cultural values. Technologies are not neutral reflections of objective reality, they are constructed within specific contexts.

Applying this lens to AI models raises some important questions:

• Who defines the rules in rule-based systems?
• Whose data trains MLMs?
• Whose data language dominated NLP systems?
• Who benefits from automation and who is harmed?

Scholars like Langdon Winner have argued that artifacts can have politics, meaning technologies can reinforce or challenge social hierarchies. AI models are powerful contemporary examples of this idea. They embed assumptions, encode classifications, and influence decisions about employment, education, healthcare, policing, and communication.

The Six Types of AI Models

• Rule Based AI systems

Rule-based AI systems operate using explicit “if-then” instructions programmed by humans. Early expert systems in medicine and finance relied on predefined logic trees to make decisions.

From an STS perspective rule-based systems make their assumptions visible. Because rules are explicitly written, we can examine who created them and what values they reflect. However, they are also rigid and may fail to adapt to complex real-world contexts. Their transparency does not guarantee fairness, it simply makes bias easier to locate.

• Machine Learning Models

Machine learning models (MLM) learn patterns from data rather than following fixed rules. Supervised learning uses labeled datasets, unsupervised learning identifies patterns independently, and reinforcement learning optimizes decisions through feedback.

ML systems are often perceived as more “objective” because they rely on data. However scholars like Cathy O’Neil in Weapons of Math Destruction demonstrate how algorithmic systems can reproduce inequality at scale. When historical data reflects discrimination, MLMs can automate and legitimize those patterns.

Similarly, Virginia Eubanks in Automating Inequality shows how algorithmic systems disproportionately impact low-income communities. Thus, machine learning models are not neutral and instead are shaped by the data infrastructures and social inequalities that they emerge from.

• Deep Learning Models

Deep learning is a subset of machine learning that uses neual networks with many layers to process large-scale data. These models power speech recognition, autonomous systems and advanced image analysis.

Deep learning systems are often described as “black boxes” because of their internal decision processes are difficult to interpret. Kate Crawford in Atlas of AI emphasizes that AI systems are built on extractive supply chains involving data, labor, and natural resources. The computational scale required for deep learning centralizes power in corporations and governments with access to massive datasets and energy resources.

The opacity of deep learning raises concerns about accountability. If a model denies a loan or misidentifies a suspect, who is responsible?

A visualization of the types of AI models used (The 6 Types of AI Models By Dr. Luke Soon)

• Natural Language Processing (NLP) Models

NLP models process and generate human language. They power translation systems, chatbots, sentiment analysis tools, and large language models.

Language is deepy cultural. Scholars such as Emily M. Bender have argues that large language models can create fluent but misleading outputs, raising concerns about misinformation. NLP systems are also disproportionately trained on English-language data, marginalizing less represented languages and knowledge systems.

Language technologies shape public discourse, education, journalism, and knowledge production. Therefore, NLP models influence how societies define truth, expertise and authority.

• Computer Vision Models

Computer vision systems analyze and interpret visual data, including facial recognition and medical imaging.

Research by Joy Buolamwini demonstrates that facial recognition systems often perform less accurately on darker-skinned and female faces. This reveals how dataset composition affects outcomes. Ruha Benjamin argues that such systems can reinforce racial hierarchies under the guise of technical objectivity.

Computer vision is frequently deployed in surveillance contexts, raising concerns about privacy, civil liberties, and state powers.

• Generative AI Models

Generative AI models produce new content, such as text, images, music, or even code, based on training data patterns. These systems have transformed creative industries and education.

Generative AI raises questions about authorship, labor, and originality. If a model is trained on millions of artworks or texts, what obligations exist to original creators? Scholars examining “ghost work”, such as Mary L. Gray, highlights the invisibile human labor involved in data labeling and content moderation.

Generative AI also challenges trust in media ecosystems by making synthetic content indistinguishable from human-produced material.

Missing Voices and Marginalized Perspectives

Across all six AI model types, certain voices are consistently underrepresented:

• Indigenous knowledge systems
• Global South perspectives
• Non-dominant language communities
• Disabled users
• Data laborers

Safiya Umoja Noble demonstrates how search algorithms heavily benefit dominant cultural narratives. Catherine D’Ignazio and Lauren Klein advocate for feminist data practices that prioritize quity and participation.

When AI models are built primarily in wealthy, western institutions, they risk exporting particular worldviews globally. Addressing missing voices requires participatory design, inclusive datasets, and meaningful community governance.

Ethical, Political, and Social Implications ACross Models

Although each AI model type has distinct technical characteristics, several cross-cutting concerns emerge:

• Bias and discrimination
• Transparency and explainability
• Environmental sustainability
• Labor exploitation
• Concentration of power

Ai systems shape access to housing, employment, education, healthcare, and public services. As such, they are not simply engineering achievements but political infrastructures.

And STS approach encourages students to ask not only “How does this model work?” but also:

• Who designed it?
• Who benefits?
• Who is excluded?
• Who is accountable?

Conclusion: AI models as Socio-Technical Systems

Understanding Jungco’s six AI model types provides a useful technical foundation. However, examining them through STS reveals that AI models are deeply embedded in social, economic, and political systems.

AI does not simply reflect society, it actually shapes it. Each model type encodes assumptions, redistributes authority, and influences human decision making. By studying AI as a socio-technical system, students can move beyond technological determinism and toward critical engagement with the future of intelligent systems

References:

Benjamin, R. (2019). Race after technology: Abolitionist tools for the new Jim code. Polity Press.

Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610–623). Association for Computing Machinery.

Buolamwini, J., & Gebru, T. (2018). Gender shades: Intersectional accuracy disparities in commercial gender classification. In Proceedings of the 1st Conference on Fairness, Accountability and Transparency (pp. 77–91). PMLR.

Crawford, K. (2021). Atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.

D’Ignazio, C., & Klein, L. F. (2020). Data feminism. MIT Press.

Eubanks, V. (2018). Automating inequality: How high-tech tools profile, police, and punish the poor. St. Martin’s Press.

Gray, M. L., & Suri, S. (2019). Ghost work: How to stop Silicon Valley from building a new global underclass. Houghton Mifflin Harcourt.

Jungco, K. (2024). Types of AI models: A deep dive with examples. eWeek.

Kukla, A. (2000). Social constructivism and the philosophy of science. Routledge.

O’Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy. Crown.

Winner, L. (1980). Do artifacts have politics? Daedalus, 109(1), 121–136.

Genesis human Experience. (2025). (L. Soon, Ed.). https://genesishumanexperience.com/wp-content/uploads/2025/09/54d6be30-bf8d-4c3c-ac22-58bb45621dae.png

AI Acknowledgement Statement:

Ai was used to create an outline and help find reputable sources for types of ai that fits the goals of this textbook chapter. The outline helped organize my information while the sources gave me a good reference on the different types of ai. I had then used further sources to back up that data. OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat