AI, short for Artificial Intelligence, has been taking the world by storm since the launch of OpenAI’s ChatGPT in 2022. Since then the availability and accessibility of bots, programs, and technologies that are powered by a technology that can be described as artificial intelligence has grown exponentially. Despite these developments there are a lot of misunderstandings about what artificial intelligence actually is and what it is capable of. Partly, this is owed to the extremely fast development of the technology, which makes it hard to keep up. This chapter therefore tries to offer an introductory definition of artificial intelligence as well as an overview of resources to keep up with recent developments and resources useful for journalistic work.
What is Artificial Intelligence?
Since the technology is changing so fast it is hard to come up with one definition that fits all systems that are somehow powered or linked to artificial intelligence. What is however important, is clearing up the misconception that AI means intelligent and creative thinking, alike to a human brain. While AI definitely has something to do with intelligent behavior there is much less “real thinking” than people might presume. AI rather describes the science of training machines, computers and algorithms, to observe human behavior and learn to imitate it as closely as possible.
Autonomous systems for example are trained and engineered to solve problems on their own and act independently based on the information and the rules they were provided beforehand.
An extremely important part of AI technology is machine learning (ML), which refers to the field that trains computer agents, mostly algorithms, to recognize rules and pattern within the data that the systems are provided and draw conclusions from that data that are as close as possible to natural human behavior and the reality.
There is also supervised and unsupervised learning where an algorithm is trained to recognize and predict human-made categories such as types of cancer or other illnesses. In cases of supervised learning the AI is provided with a lot of images for example of different kinds of skin cancers and based on these labeled images learns to recognize the type of cancer when provided with new images. Unsupervised learning means that the algorithm learns to make its own predictions about human behavior based on the data it was previously provided. Reinforced learning refers to a computer program being trained to combine different action sequences in such a way that the reward or the outcome is highest based on the data on the actions and rewards that was provided beforehand (for example chess computers). However, the program here draws its own conclusion based on the data and is not informed about what by human standards is considered the best behavior.
Related to these different kinds of learning methods is the field of natural language processing (NLP), which refers to the processing and imitation of language in audio and text. Based on a huge amount of data on human language as well as results from linguistics, NLP aims to develop a technology that is able to process speech, recognize different kinds of texts, understand the meaning of bodies of natural language and create its own output of language that is correct grammatically and content wise.
Lastly, it is important to highlight that AI is not just a sophisticated algorithm. An algorithm is a computer program that contains a step-by-step instruction on how to solve specific problems and how to behave in specific situations. The more sophisticated the algorithm the more possible behaviors for certain situations it contains. AI systems are often based on algorithms but what makes them “intelligent” is that they develop solutions and rules based on the data and feedback that they are provided: for example the AI trained to detect skin cancer is not provided with the list of rules doctors learn for detecting skin cancer but is merely provided with the feedback whether or not an image shows skin cancer. From all the information in that image, and all the other images fed into the data base, the AI then develops its own predictors for skin cancer that are refined and adjusted depending on the feedback provided to the AI.
What Makes AI an Ambivalent Technology?
There are a lot of problems and open questions surrounding this new technology so this chapter is only going to describe the most obvious ones. AI can inherit biases from the data the systems are trained on, leading to unfair or discriminatory outcomes. Privacy is a concern as AI systems collect and analyze vast amounts of personal information, sometimes without consent. There are also a lot of unsolved copyright problems around this issue because a lot of the big natural language most likely had materials in the data their AI was trained on that was protected by copyright laws. There are several pending lawsuits that will ultimately determine how courts and states deal with this problem.
Furthermore many AI models are “black boxes,” making their decision-making processes opaque and difficult to trust. They can also generate misinformation, such as deepfakes, which can manipulate public opinion or harm individuals. Ethical dilemmas arise over responsibility for AI decisions and its potential misuse, such as in autonomous weapons. Additionally, training large AI systems consumes significant energy, impacting the environment, and the huge amount of energy needed to run AI systems as well as the computer and network technology needed to access these systems creates global inequality when it comes to access of the technology.
How Can AI be Used in the Newsroom?
There are countless ways that AI technology can be used in the newsroom. As previously stated it is a big and fast developing field and there are different technologies that journalists can utilize in different steps of their workflow. From building topic specific chatbots and automated factchecking to summarizing documents for research and the creation of (semi-)automated contend – there are a lot of possibilities. If you want to learn more about the topic, here are a view great resources for journalists:
- This article by Medium is a great introduction for non-experts which dives a little deeper into the technological and historical development of the technology.
- If you are already familiar with the basics of the technology and want to expand your knowledge on the use of AI for news organizations, Charlie Beckett’s articles on Medium are a great starting point. Beckett is a professor for journalism at the London School of Economics and has dedicated the past few years of his research to the question how AI is going to change journalism. Following him on Bluesky can also be a way to keep track of new developments in the field.
- Kuek Ser Kuang Keng is a Malaysia-based data journalist that works at the Pulitzer Center and is currently working on covering the cross-roads of data journalism and algorithm accountability.
- The Reuters Institute is also a great source of information, since they report on new developments on AI in journalism but also do research on the impact of the technology on news room and audiences’ response to different AI-powered technologies.
- Lastly, JournalismAI is a initiative from the London School of Economics and the Google News initiative that not only offers a newsletter about new developments in the field but also offer fellowships, conferences, video sessions on different aspects of AI-development as well as a regularly updated and very extensive introduction to the field of artificial intelligence. And even better: together with the Google News initiative hey offer a free online course that provides journalist with extensive and easily accessible training on AI for the newsroom:
Since the field of artificial intelligence is currently developing so fast, it is hard to provide an introduction to the topic on a web-manual that is updated only every few years. This guide is supposed to serve as a first introduction to the topic and provide inspiration as well as resources for everyone interested to learn more about artificial intelligence. As always, there is a lot more out there and the resources listed here best serve as first stepping stones on jour journey to learn for about AI and its many roles in the newsroom.