Article Text

Download PDFPDF
Pathology training in the age of artificial intelligence
  1. Ananya Arora1,
  2. Anmol Arora2
  1. 1Selwyn College, University of Cambridge, Cambridge, Cambridgeshire, UK
  2. 2School of Clinical Medicine, University of Cambridge, Cambridge, UK
  1. Correspondence to Anmol Arora, School of Clinical Medicine, University of Cambridge, Cambridge CB2 0SP, UK; aa957{at}cam.ac.uk

Statistics from Altmetric.com

Introduction

There are two key emerging technologies which are anticipated to transform healthcare in the next 10 years: artificial intelligence (AI) and robotics. Of the two, AI has attracted particular interest due to scope for generativity and promising results from early studies into its potential implementation. Indeed, a recent study suggested that approximately 80% of pathologists believe that AI will become integrated in diagnostic workflows in the next decade.1 There is scope for both robotics and AI to transform pathology as a speciality in the near future, though the two innovations are not mutually exclusive. Simple robots which are narrowly programmed to perform some physical actions already exist, but to create advanced robots would require synergies with AI. Before we see robots take over the physical actions of pathologists, we are likely to see AI have significant impact in other ways. This can partly be attributed to Moravec’s paradox, an observation by AI researchers that, to program AI which is capable of advanced cognitive processes is often relatively straightforward compared with simple physical tasks.2 Pathology as a speciality is particularly pertinent to emerging AI research, which currently focusses on image and data analysis, two key elements of a pathologist’s role. Early research has begun to explore how AI may begin to affect pathology and improve patient care but the effects on pathology training remain relatively underexamined.

Types of artificial intelligence

Most research into AI has focussed on deductive systems. However, there are other types of AI which do not attract as much research or media speculation (table 1). Broadly speaking, there are four categories:

View this table:
Table 1

Comparison of different types of AI

  • Deductive AI

  • Generative AI

  • AI for workflow optimisation

  • AI in robotics

Deductive systems function by analysing data sets and finding patterns which would be infeasible for humans to program. Their uses are well …

View Full Text

Footnotes

  • Handling editor Runjan Chetty.

  • Twitter @AnmolArora_98

  • Contributors Both authors contributed to the writing of this Viewpoint article and the ideas expressed within it.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; internally peer-reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.