HCI - Lecture 19: Current Issues in HCI Part 1 HCI Stephanites et al. (2019) notes that there has been a fundamental change in HCI in recent years, and that technology is becoming more truly interactive · Artificial intelligence for example . Although noted by Kaplan (2016), there are a lot of misunderstandings and misconceptions around artificial intelligence, even within the field Big Data Multiple definitions, including: "Data that is so extensive it could not be stored, processed, shared and analysed using conventional means in a reasonable timescale" However, context matters - e.g. a 100mb file is too large to send by email, and so in those circumstances could be referred to as big data as it cannot be shared using conventional means Landers et al. (2016) argue that big data has several advantages to data collected through more traditional means in psychology · The data is primarily behavioural, instead of self-report data from surveys and questionnaires . The sample size that can be obtained is substantially larger than that which can be typically achieved in psychology studies . The people from whom the data is being collected are not aware of this, and as such issues such as demand characteristics are reduced · Data can also be collected at a speed that would not normally be possible · Data collection is possible in populations where there may otherwise be logistical and practical challenges, such as in developing nations Social Media It has been reported that Facebook alone generates 4 new petabytes of data every day · This is the equivalent of 2000 billion pages of printed text Opportunities for psychologists? · Collecting data is often one of the biggest challenges in psychology This is a significant example of how big data can provide psychologist with research data in a way that would have been hard to imagine 20 years ago . It also illustrates the extent and value of the information that social media platforms hold their users
Machine Learning and AI Humans learn through past experience Machines, however, typically do not learn, and instead follow a set of pre-set instructions . It will follow these instructions exactly, even if the desired outcome is not being achieved . And will do the same thing again the next time, unless it's modified by a human One of the aims of machine learning is to develop software and devices that learn from their previous experience Machine learning can be coupled with big data, with software learning quickly from data that would be far too large and complex for a human to process · Any consequences of this? . Machine learning can lead to systems that behave in ways that are highly effective, but which no human can understand or explain It's also possible for machine learning and big data to be used to predict future behaviour . The more data the more precise the prediction . Example: would you be able to predict if someone would like a song they have