Table 1.
Brief Description of the Key Terms Used in This Review Explaining AI, Ethics, and Pathology
| Terminology | Explanation |
|---|---|
| Artificial intelligence | Intelligence emulated or simulated by the use of technological means. Computational machinery is used to achieve intellectual autonomy and independence of thought similar to that seen in humans |
| Algorithm | A set of step-wise commands to accomplish a specific task/goal/objective. In AI, algorithms are the programming code that enable the functionality of an objective task and are key to emulating intelligence in an artificial manner |
| Bias | Discrimination in favor or against a set of outcomes in a particular setting. In AI ethics, this often refers to the ability of the AI algorithm to discriminate against individuals, groups, or populations based on the design of the original algorithm |
| Big data | Big data refers to data produced by an automated and repetitive technological process. Big data may be quantified in terms of the abundance of the data size generated. In pathology, some examples of big data are a digital pathology whole slide image repository, or databases containing complete blood counts across a population and time |
| Data privacy | The moral, legal, and ethical expectations to maintain confidentiality of data collected from either individuals or non-individual resources. In pathology, institutions responsible for the collection of patient laboratory data are tasked with the responsibility to ensure data privacy at individual and population levels |
| Data-shifts | A concept referring to the change in the data distribution between training and real-world data sets in AI algorithm development |
| Digital pathology | An emerging paradigm of pathology focused on digitization of traditional glass-based slides read by pathologists. Digitized slide data can be stored, viewed, and shared in real time, leading to enhanced efficiency of the sign-out process |
| Ethics | A branch of philosophy studying the concepts of right and wrong human behavior in a systematized manner |
| Machine learning | Computational algorithms that are capable of automated learning processes through iterative feedback of data without (or with minimal) human intervention |
| Underspecification | Failure to specify adequate details in the context of a training set of an AI algorithm |
AI, artificial intelligence.