Believe it or not, you’ve most likely used Artificial Intelligence (AI) at least once in your life, if not every day. Just think, face ID, Siri, Google Maps, and social media are all powered by AI technology. Artificial intelligence (AI) is a technology that allows machines or software to perform human-like tasks. There are six subsets of artificial intelligence, specifically machine learning, deep learning, robotics, neural networks, genetic algorithms, and natural learning processing (Keserer, 2023).
Machine learning is the concept of machine learning human behavior from data and algorithms and improving over time(). Some of the most common examples of machine learning are the Google search engine, TV streaming apps (i.e. Hulu, Netflix), and online businesses (i.e.Target, Amazon). Further, deep learning is another subset of artificial intelligence as well as machine learning. Deep learning utilizes neural networks to learn how to perform tasks and helps to recognize patterns (Keserer, 2023). Self-driving cars, translation apps, chatbots, and facial recognition all apply deep learning to achieve these actions.
Besides machine learning and deep learning, there is robotics, which is a physical form of artificial intelligence (Keserer, 2023). Robotics are used in multiple different lines of work and are responsible for performing tasks that are deemed dangerous or hard. An example of a robot includes a robotic exoskeleton, a medical device worn by a human recently recovering from an illness or injury. Another example is an unmanned combat aerial vehicle (UCAV) or a combat drone. A UCAV is responsible for surveillance, detecting targets, carrying weapons, and more. Equally important there is neural networks, which is a type of software that is used to recognize patterns (Keserer, 2023). Further, the patterns are used to help AI process information and make decisions like humans (Keserer, 2023). However, one of the issues associated with neural networks is that the software overgeneralizes and is limited in its ability to interpret data.
Apart from that, there are genetic algorithms, which use genetics to solve issues and follow the process of natural selection. Genetic algorithms utilize four steps, initialization, evaluation, selection, and reproduction (Keserer, 2023). An example of genetic algorithms would include altering DNA to present the “fittest” of a species. Lastly, there is natural learning processing, which enables machines to analyze, change, and respond to human language (Keserer, 2023). Translation apps, Alexa, Siri, search engines, and Grammarly all use AI technology, specifically natural learning processing. In essence, AI is more complicated and used more in your daily life than you may have originally thought.
Keserer, E. (2023, November 24). The six main subsets of AI: (Machine Learning, NLP, and more). Akkio. https://www.akkio.com/post/the-five-main-subsets-of-ai-machine-learning-nlp-and-more
I really enjoyed your post on the widespread use of AI. Frequently we do not even realize that it stands behind common tasks that we do every day. While writing this post I enjoy having a spell checker and grammar correction available instantly without thinking much about it. Artificial Intelligence can take many shapes and forms, the most recent one was made popular by ChatGPT, a conversational chatbot from OpenAI. As any new technology it brings both upsides and downsides. One of the downsides specific to social science research was reported recently which caught my attention.
Historically, a lot of research in psychology and other social sciences has been done by using university students for surveys, questionnaires, etc. While students are readily available there is a danger of a bias towards younger, somewhat better educated part of the population. Crowd-sourcing platforms such as Amazon Mechanical Turk brought a solution for the problem of finding a large diverse group of people to engage with for social research (Samuel, 2018). It is easy, convenient, and fairly cheap to use. A researcher can get thousands of replies within minutes for a survey.
Appearance of generative AI such as ChatGPT changed the situation dramatically. Apparently up to 46% of mechanical turk workers are using ChatGPT or other forms of Large Language Model to automate their work. This presents a danger for social science research as the replies to the surveys are now coming from a chatbot instead of real people. This makes any output of such study unreliable or completely made-up. One more thing to worry about when you are conducting applied psychology research.
References
Coldewey, D. (2023, June 14). Mechanical Turk workers are using AI to automate being human. TechCrunch. https://techcrunch.com/2023/06/14/mechanical-turk-workers-are-using-ai-to-automate-being-human/
Samuel, A. (2018, May 15). Amazon?s Mechanical Turk has Reinvented Research. JSTOR Daily. https://daily.jstor.org/amazons-mechanical-turk-has-reinvented-research/