AI and Human Factors'
In this post, I will give a brief overview of how the areas of Artificial Intelligence (AI) and Human Factors (HF) overlap along with the ways HF can be used to improve and guide the development and applications of AI. The key area where they interact with each other in the field of Human-Computer Interaction (HCI). AI is increasingly supporting high-consequence human decisions from healthcare to criminal justice which depending on how particular AI systems were trained or designed through Machine Learning (ML) can have serious consequences. These are often in the form of bias. HF has the potential to identify and reduce instances of these bias and related problems.
Applying Human Factors to AI
Increasing machine, intelligence leads to a shift from a mere, interactive to a much more complex cooperative human-machine, relation requiring, a multidisciplinary development, approach were
AI frame opportunities and challenges for user interface design. Reducing bias' through ensuring testing of systems doesn't misidentify or preference certain people over others. Auditing and verifying the type of data the AI was trained on. Also, what type and kinds of data point the AI/ML is using to draw its conclusions. For example, was the AI allowed to train itself with no guidance or was it programmed to focus on key data points as selected by the designers themselves.
Principles for human-AI interaction have been discussed in the human-computer interaction community for over two decades, but more study and innovation are needed in light of advances in AI and the growing uses of AI technologies in human-facing applications. HCI can be said to be a topic that naturally arose with the birth of the computer. Its scientific origin can be traced back to 1960sarticle published by JC Licklider called "Man-Computer Symbiosis", which mentions that humans should interact with computers and collaborate to complete tasks. However, in the era of artificial intelligence in full swing, related issues in human-computer interactions seemed insignificant. For example, how to optimise interface layout, command names, and text editors. People are more willing to believe that the upcoming intelligent machines will solve all problems including these problems. It has been noted that HCI and AI follow alternative ups and downs in improvements.
In the end, anything an AI does has to fit into a human-centred value system that takes our unique human abilities into account. HF has the potential to improve user interaction with AI and its applications. Hopefully, this post provides a useful overview of AI and HF. It is quite an in-depth topic and if you are interested there is a fair bit of further reading on the topic.