Literature Review: HCI Theories and Contributions to AI Systems

This report presents a literature review that explores the significant relationship between Human-Computer Interaction (HCI) and Artificial Intelligence (AI). It begins by defining HCI and its theoretical underpinnings, including Fitts's Law, the Keystroke-level model (KLM), GOMS, and the Model Human Processor, detailing how these theories provide a framework for understanding human interaction with computer systems. The review then shifts to examine the contributions of HCI to the development of AI systems, highlighting the convergence of the two fields and the role of HCI in advancing Machine Learning techniques, processing large datasets, and incorporating the human component in AI system design. The report emphasizes the importance of HCI in addressing issues such as algorithm transparency and improving the interoperability of AI systems, ultimately advocating for the integration of human intelligence as a core component in AI development.

Topic: Literature Review in HCI
Name
Institution

Theories used in HCI
HCI theories provide a through a pedagogical survey of HCI science. HCI is the
intersection between computer and information technology on one and social and behavioral
science on the other hand. A theory in HCI is an attempt to explain reality. Usually, a theory is
based on a model thus it must be disprovable through experiments. There are various theories in
HCI. One of the theories is by Paul Fitt. The theory is an empirical model that describes speed-
accuracy trade-off features of human muscle drive with an analogy of Shannon’s channel
capacity. Fitts’ theory predicts how long it takes for a user to select a target used for evaluating
device input. Paul Fitts, the author of Fitts’ theory, expanded the research by Woodworth’s in
HCI that focused on the performance of a telegraph operator. The parameters of Fitts’ law are the
movement distance from the starting or the initial position to the target center, the time to move
to the goal and target width (Boritz, s, & Cowan, n.d).
Second is the Keystroke-level model (KLM) HCI theory. KLM theory is a description of
users’ tasks which are based on low-level activities or actions. The theory foresees how long a
user will take to complete a routine task without experiencing any errors or mistakes using what
Allen referred to as an interactive computer system (Nicole C. Krämer, 2012). The theory was
proposed by Stuart, Allen, and Thomas in 1980. The theory suggests six operators where one of
the last operators is a mental operator, another one is the response operator and the first four are
physical motor operators. There is the K operator which is the most frequent operator, then there
is the P operator, which points to the target, there is the H operator which includes the measure
between two devices and the fine positioning of the hand, the D operator, and the M operator
(Nardi, n.d).

There is the GOMS HCI theory; it is a higher level KLM with hierarchy and structure.
GOMS (Goals, Operators, Methods, and Selectors) is a group of those models which are
predictive. These models are used to improve the efficiency of a computer machine and that
interaction between humans and machines. The model achieves this by eliminating and
identifying redundant user actions. The theory is widely used by usability specialists as it
produces both qualitative and quantitative predictions of individuals with using the proposed
system (Juan Cruz-Benitoa, 2018).
Model Human Processor theory of HCI was developed by Newel, Card, and Moran in
1983. The theory attempted to draw that analogy between the ways humans perceive processes,
remember things and the way they process things. This would then allow HCI designers to
predict the types of HCI interface. The theory focuses on three major processes which are
perceptual store, long term memory, and short term memory. The theory tries to calculate how
long it can take to perform a certain task (CAI, WINTER, STEINER, WILCOX, & TERRY,
2019).
How HCI contributed to AI systems
According to Jonathan Grudin, AI and HCI are converging. Initially, HCI was seen as a
methodology that focused on improving applications while AI methodology focused on future
possibilities. Both HCI and AI focuses on economic and intellectual resources. Currently, most
of AI researchers are working with widely available systems developed using HCI theory and
have greatly appreciated the need for usability when developing AI systems (McKilligan, 2018).

HCI has undergone great developments and scientists in this field are trying to bridge that
gap between machine and man. These efforts have led to not to the development of touch
screens, mice, and keyboards but also the development of AI systems. According to Kori (2019),
deep learning advancement, as well as advances in hardware, have elevated or pushed the
development of artificial intelligence systems. HCI has contributed to this by assisting in the
development of Machine Learning techniques which in turn assisted in the processing of large
volumes and data modalities. Though HCI assisted in turning voluminous sources of data into
signals. It also assisted in supplementing AI systems by replacing human making decisions.
Through HCI, AI systems began to make great strides in lots of issues of societal significance; in
fact, it has led to a lot of contributions in development, education, agriculture, environment, and
healthcare. Also, HCI has assisted AI by providing mathematical insights to correct issues like
the transparency of algorithm choice. With AI researchers are now able to improve on those
algorithms thus able to improve interoperability in HCI systems. Lastly, HCI has led to the
realization of the human component in the development of AI systems. In any AI system
development, the human component is being advocated for either as augmentation for human
intelligence or as an assistant (Inkpen, Veale, Chancellor, & Baumer, 2019).

References
Boritz, J., s, K., & Cowan, B. W. (n.d). The Glossary of Human Computer Interaction. Graphics
interface' 19, 216-223. Retrieved from Interaction Design Foundation:
https://www.interaction-design.org/literature/book/the-glossary-of-human-computer-
interaction/fitts-s-law
CAI, C. J., WINTER, S., STEINER, D., WILCOX, L., & TERRY, M. (2019).
“HelloAI”:UncoveringtheOnboardingNeedsofMedical PractitionersforHuman–
AICollaborativeDecision-Makin. Proc. ACMHum.-Comput. Interact., 104.
Inkpen, K., Veale, M., Chancellor, S., & Baumer, E. P. (2019). Where is the Human? Bridging
theGap Between AI and HCI. CHI 2019 Workshop Summary (pp. 1-8). Glasgow.
Scotland: Glasgow Press.
Juan Cruz-Benitoa, . F.-P. (2018). Analyzing the software architectures supporting HCI/HMI⁎
processes through a systematic review of the literature . Telematics and Informatics, 118-
132.
McKilligan, H. P. (2018). A Systematic Literature Review for Human-Computer Interaction and
Design Thinking Process Integration. HCI and DT Process Integration , 726-737.
Nardi, B. A. (n.d). Activity Theory and Human-Computer Interaction. In K. Kuutti, Activity
Theory as a potential framework for humancomputer interaction research (pp. 103-116).
NY: MIT press.
Nicole C. Krämer, A. M. (2012). Human-Agent and Human-Robot Interaction. In N. C. Eimler,
Human-Agent and Human-Robot Interaction Theory: Similarities to and Differences

from Human-Human Interaction (pp. 215-216). NY: Elsevier Press. Retrieved from
Berkeley: http://blogs.ischool.berkeley.edu/i213s12/files/2012/01/i213-13.pdf