It is 2005, and the first page of Jane’s Facebook feed still showed Uncle Osasu’s 10:25 am prayer devotional post in the evening. She has had fewer than three new posts by 7 pm that night. Unbelievable in this age of information overload, but yes, this was the case in the early days of social media. Posts were primarily displayed in chronological order, showcasing the latest updates from friends as they were being shared. It was a time when the digital landscape was less cluttered, and the experience felt more organic, highlighting the most current interactions without the influence of algorithms, which shifted users’ experience on social media platforms.
As the digital space continues to evolve, algorithms have become essential for social media platforms to filter, regulate, and promote content. The BBC reported Facebook as one of the pioneers of recommendation algorithms on social media, having one of the largest feeds. The system prioritises posts based on various factors not just the time they were posted, allowing older posts to resurface depending on user engagement and preferences.
The algorithm also gives people control over what they see by allowing them to unfollow pages, report content, indicate “interested” or “not interested” on the posts they see, etc.
What Algorithm Means
An algorithm is a set of instructions designed to perform a task automatically. With respect to social media, Parida Surabhi described it as being “responsible for maintaining order, ranking search results, and ensuring that feeds remain organised and relevant.”
Originally implemented by Facebook in 2009 with ranking metrics and personalised feeds, other digital platforms like Twitter (X) and Instagram soon adopted similar approaches.
With AI, algorithms now extend beyond simple ranking and personalisation to actively recommending content based on observed user patterns.
This ability makes AI-powered algorithms information gatekeepers, managing and regulating content delivery to users.
While algorithms provide the foundation for managing and displaying content on social media platforms, AI supercharges this process by enabling smarter, data-driven decisions that improve user engagement and satisfaction.
It’s not just social media platforms that benefit from AI enhancements. Early search engines used algorithms for indexing and keyword-based searches, but with the integration of AI, particularly through machine learning, deep learning, and natural language processing, search engines now do more than retrieve data based on keywords; they can understand complex inputs and even generate output like the Google Gemini and other AI-powered browser extensions.
AI-powered algorithms gained momentum with the surge of big data in the 2010s, following the popularity of social media, early internet and smartphones, and with it came benefits, including improving user experience by allowing users to connect to relevant content based on their previous searches and engagements, and even follow up on similar topics or handles. It has also helped small and medium-scale businesses, especially those without physical shops, to promote their business and retain engagement. Danielle Draper of Partisan Policy also noted that social platforms can restrict illegal, harmful, and inappropriate content on these algorithmic systems, protecting children and safeguarding marginalised groups.
However, AI-driven algorithms are susceptible to amplifying hate speech, threats, cyberbullying, and false narratives. There are concerns about how users can escape once trapped in a web of information through personalised feeds.
A previous article highlighted how these personalisations, based on users’ behaviour, limit their encounter with other viewpoints and how disinformation actors exploit the algorithm to promote misinformation, using clickbait titles and emotionally charged imagery to gain visibility and trap readers on the web.
How AI Algorithms Act as Gatekeepers
AI-driven algorithms now determine what information and content individuals are exposed to online, acting as unseen gatekeepers that influence perception and engagement. Some of the ways it plays this role include:
- Content recommendations: This is when content is recommended to users based on their information consumption history. ML, Deep learning and Natural language processing enable this process.
- Content Moderation: Through the enforcement of community guidelines, AI algorithms are used to detect and remove content that violates social media platforms’ policies.
- Content Generation: This is exemplified by the recent integration of Google Gemini into the Google search engine. Gemini output is the first result that pops up, showing you insights from sources that demonstrate Experience, Expertise, Authoritativeness and Trustworthiness (E-E-A-T).
There are also other web extensions that generate text output from user search results, like Monica. These AI-driven algorithms are also shifting traditional SEO practices by using users’ behaviour patterns, content, and intent to filter content to deliver personalised responses.
- Engagement-Driven Metrics: Many AI algorithms are optimised for engagement metrics such as likes, shares, comments, and time spent on content. Sensational, emotionally charged, and even false information often garners higher engagement and provokes stronger reactions. Algorithms, in their pursuit of maximising these metrics, can inadvertently prioritise and push such content to a wider audience, even if it is misleading or harmful.
- Speed and Scale of Dissemination: AI allows for the instantaneous analysis and dissemination of vast amounts of information. This speed, while beneficial for breaking news, also means that false or misleading content can spread globally at an unprecedented rate before human fact-checkers can verify and debunk it. Automated bots and coordinated networks, often powered by AI, can further amplify this spread, making it difficult to contain.
Should AI Algorithms be Gatekeepers
Traditionally, individuals/institutions such as journalists, factcheckers, researchers, etc., not only contributed to the body of knowledge, but also served as gatekeepers, making the decision of what information gets to the public and how these issues are framed. These decisions are not based on criteria such as the interest of the audience alone; gatekeepers such as newsrooms/journalists, for example, subscribe to a code of ethics that guides the profession. In this way, these gatekeepers shaped public discourse. As the internet and social media reshaped information access and distribution, these traditional gatekeepers are no longer central to information distribution and narrative framing. According to Nicholas Barrett, a technology reporter at the BBC, social media algorithms have shaped the flow of information, allowing for a more diverse range of voices and viewpoints to emerge in journalism and information sharing.
Now, while traditional gatekeepers were not infallible, there are systems of accountability that act as safeguards. No such systems exist for algorithmic gatekeepers, which has made them significant drivers of misinformation and disinformation. The big questions across the world are, who bears the burden of responsibility for this new algorithm-driven world order? How do we design an accountability structure? Who builds and monitors these structures? There is as yet no concrete answer to these questions.
The World Economic Forum’s 2024 Global Risk Report identified misinformation as the most significant threat over the next two years. Media expert Prof. Stan Karanasios from the University of Queensland elaborated on this, stating that algorithms pushing AI-generated falsehoods are exacerbating the problem of misinformation, disinformation, and fake news through automated content generation, which is pushed by Algorithms.
While there are clear benefits to all the technological advancements that have brought the world, AI Chatbots have revolutionised research, turning a 5-day stay in the musky library with bad, expensive coffee into a 1-hour sprint from the comfort of a toilet seat. However, its role in controlling online information flows poses a significant risk for information disorder. The speed and ease with which these systems thrive is their Achilles heel, bringing completely fabricated data and citing non-existent sources framed in convincing language in quick time.
Efforts such as community notes on X, community guidelines, policy agreement clause and even collaboration between big tech and the fact-checking community are some of the attempts to keep up with the foibles of an algorithm-driven information ecosystem. But these are struggling to keep up, weakened by inconsistent policy too often influenced by private interests and political headwinds, compounded by insufficient financial support.
The pursuit of user engagement, the formation of filter bubbles, the rapid spread of information, and the growing sophistication of AI-generated content all contribute to a challenging environment where even above-average users struggle to differentiate truth from falsehood.
While the questions about checks and balances of algorithmic dominance over information ecosystems will continue, the technology landscape continues to evolve, as does the role of AI gatekeeping.
Answering these questions and responding to the challenges the new information world order presents will require continuous multi-pronged approaches from policy making, to laws and precedents, involving all stakeholders from big tech to governments, and all the actors in between.
The issues to be addressed are clear: responsible AI development, algorithmic transparency, media literacy, and robust fact-checking mechanisms to complement existing efforts aimed at ensuring that AI is a tool for development rather than a catalyst for mayhem. How they will be addressed is the grey area.
