Human and Artificial Cognition: A Review of Siemens et al. (2022)

The rapid advancements in artificial intelligence (AI) have sparked debates about the timelines for achieving artificial general intelligence (AGI) or even superintelligence. However, Siemens et al. (2022) argue that for researchers and practitioners in education, these long-term predictions are secondary to the more pressing issue of understanding how AI, as it exists today, impacts human cognition and knowledge practices like learning, sensemaking, and decision-making.

Their paper, Human and Artificial Cognition, emphasizes the immediate need to explore the functional dynamics of human-machine interactions, treating human cognition and artificial cognition (AC) as distinct yet complementary systems. This blog post reviews their compelling argument for a functional, integrated approach to understanding and leveraging these two cognitive systems.

AI: A Present Reality, Not a Future Speculation

The authors remind us that AI is no longer a futuristic concept—it actively influences our daily lives. From shaping the information we encounter online to enhancing decision-making processes in industries like healthcare and education, AI’s presence is ubiquitous. However, this pervasive influence raises critical questions about the roles AI and humans should play in cognitive tasks.

Siemens et al. adopt a functional perspective, emphasizing the need to delineate which tasks are better suited for machines and which require the nuanced judgment of human cognition. This approach lays a practical foundation for addressing how humans and machines can collaborate effectively, rather than speculating on hypothetical superintelligence.

Human and Artificial Cognition: A Functional Approach

The core argument of the paper is that human and artificial cognition operate as distinct systems with unique strengths. While humans excel in creativity, sensemaking, and emotional intelligence, machines shine in processing large datasets and performing repetitive or routine tasks.

Siemens et al. classify cognitive tasks into three dimensions:

1. Sensory Processes – Tasks involving data gathering and initial processing, where both humans and machines can collaborate. For instance, AI can gather market trends, while humans interpret the emotional or contextual significance.
2. General Operations – Routine or repetitive tasks like organizing data, which are better suited to machines for efficiency and scalability.
3. Complex Integrated Activities – Tasks requiring creativity, judgment, and emotional intelligence, such as brainstorming or evaluating solutions, which remain the domain of human cognition.

This division provides a structured framework for understanding how the strengths of human and artificial cognition can be leveraged in tandem.

Bias and Ethical Concerns in AI

One of the most significant challenges highlighted by Siemens et al. is the issue of bias in artificial cognition. They discuss how algorithms like COMPAS, used in the U.S. justice system, perpetuate systemic inequities due to biased datasets. For instance, COMPAS’ risk assessment unfairly assigned higher recidivism scores to Black individuals, a reflection of decades of biased policing rather than objective evaluation.

This example underscores the cyclical nature of human-AI interactions: human biases shape AI systems, which in turn influence human decision-making, potentially amplifying existing inequities. The authors call for ethical oversight and robust mechanisms to identify and mitigate such biases.

Practical Implications for Collaboration

The paper also explores the real-world integration of human and artificial cognition in domains such as education, healthcare, and creative industries. For instance, in creative problem-solving, machines can assist by analyzing vast datasets and generating prompts, while humans evaluate these outputs and brainstorm innovative solutions.

The authors argue for tailored approaches to collaboration based on the specific needs of different domains. In education, AI could support personalized learning by analyzing student performance data, while teachers focus on emotional and social aspects of learning. Similarly, in healthcare, AI can process patient records and predict outcomes, while human professionals provide contextual understanding and empathetic care.

Research and Future Directions

Siemens et al. emphasize the need for further research at the intersection of human and artificial cognition, particularly in:

1. Task Allocation – Understanding which tasks are best suited to humans or machines in various contexts.
2. Integration Mechanisms – Developing models to seamlessly integrate outputs from both systems.
3. Affective Dimensions – Exploring how emotional and psychological factors influence human-AI collaboration.

They also stress the importance of domain-specific research, as the nature of HAC integration will differ significantly between fields like education, military operations, and emergency management.

Conclusion

The future of knowledge work, Siemens et al. argue, lies in the seamless collaboration of human and artificial cognition. By focusing on the cognitive, rather than speculative, aspects of this collaboration, their paper provides a practical framework for researchers and practitioners to optimize human-machine interactions.

The insights from this study are invaluable for navigating the ethical, cognitive, and practical challenges of integrating AI into our knowledge ecosystems. As we continue to rely on AI in increasingly complex ways, understanding the boundaries and synergies between human and artificial cognition will be critical to unlocking its full potential.

Citation
Siemens, G., Marmolejo-Ramos, F., Gabriel, F., Medeiros, K., Marrone, R., Joksimovic, S., & de Laat, M. (2022). Human and artificial cognition. Computers and Education: Artificial Intelligence, 3, 100107. https://doi.org/10.1016/j.caeai.2022.100107