Artificial intelligence (AI) is often spoken of as if it were a single thing made by a single kind of expert. In practice an AI system is the work of a team, and each member brings a different training. This overview describes the main specialties involved, what each one does and what knowledge it rests on.
The foundation: models and data
Two roles sit at the base of almost every AI project.
Machine learning engineer
Machine learning is the branch of AI in which a program learns patterns from examples instead of following rules written by hand. The machine learning engineer designs, builds and maintains these systems. The work calls for solid programming, most often in Python, and for mathematics, in particular statistics and linear algebra.
Data scientist
A model is only as good as the data it learns from. The data scientist collects, cleans and analyses large sets of data and draws out what they show. Much of the work is careful preparation: finding errors, filling gaps and checking that the data fairly represents what it is supposed to describe. The ability to present results in clear charts is part of the job, because the findings must be understood by people who are not specialists.
Specialists in particular kinds of task
Other roles concentrate on one type of problem.
Computer vision engineer
Computer vision is concerned with systems that interpret images and video: recognising objects, reading handwriting, analysing medical scans. It draws on image processing, on deep learning (a machine learning technique based on layered neural networks) and sometimes on computer graphics.
Natural language processing engineer
Natural language processing (NLP) deals with human language: translation, speech recognition, search, and systems that answer questions or produce text. The specialty combines computer science and deep learning with linguistics.
Robotics engineer
Robotics brings AI into the physical world. A robotics engineer designs, builds and programs machines that sense their surroundings and act in them. The field joins mechanical engineering, electrical engineering and software.
Newer roles: ethics and social purpose
As AI systems come to be used in decisions that affect people, two further kinds of work have gained attention.
AI ethics specialist
This role is concerned with fairness, transparency and accountability. It asks who may be harmed by a system, whether its decisions can be explained and who answers for its mistakes. People come to it from philosophy, law and the social sciences as well as from engineering.
AI for social good
Some specialists apply AI to public problems in fields such as health care, the environment and poverty reduction. The work joins technical skill with knowledge of the field in question. Its results depend on how well the people who build a system understand the setting in which it will be used.
The wider team
Several other roles are needed in most projects.
- Software engineers build the reliable systems in which a model runs and through which people use it.
- Data analysts prepare and examine the data that feeds a model.
- Research scientists develop new methods and test their limits.
What the list shows
Three things stand out. First, the field is broad. It needs mathematicians and programmers, and also linguists, engineers of physical machines and people trained to think about law and ethics. Second, the roles depend on one another. A model built without good data, sound software or attention to its effects is of little use. Third, the boundaries move. Titles and duties differ from one employer to another, and they change as the technology changes.
For anyone trying to understand AI, whether as a student, a teacher or an interested reader, it helps to remember that it is made by people in different roles who make choices at every step. Knowing who does what is a first step towards asking informed questions about the systems that result.
