{"id":2681,"date":"2022-09-01T19:15:00","date_gmt":"2022-09-02T00:15:00","guid":{"rendered":"https:\/\/ic4ml.org\/?post_type=journal-article&p=2681"},"modified":"2022-10-27T19:13:33","modified_gmt":"2022-10-28T00:13:33","slug":"reimagining-critical-literacy-in-the-age-of-the-algorithm","status":"publish","type":"journal-article","link":"https:\/\/ic4ml.org\/es\/journal-article\/reimagining-critical-literacy-in-the-age-of-the-algorithm\/","title":{"rendered":"Reimagining Critical Literacy in the Age of the Algorithm"},"content":{"rendered":"\n
\n
\n

Abstract<\/strong><\/h2>\n\n\n\n

Algorithms are playing an ever-increasing role in the way that we understand and interact with our world. This paper provides an overview of Artificial Intelligence technologies, explores the promise of algorithms as well as the issues and considers the implications of algorithms on education and literacy practices.<\/p>\n\n\n\n

Keywords<\/strong><\/h2>\n\n\n\n

Algorithms, Literacy, AI, Artificial Intelligence, Education, Deep Learning, Bias, Algorithmic Discrimination, Media, Social Media, Data Literacy<\/p>\n<\/div>\n\n\n\n

\n
\"\"<\/figure>\n<\/div>\n<\/div>\n\n\n\n
\n\n\n\n

Algorithms have an ever-increasing impact on our daily lives on both a conscious and a subconscious level. Have you stopped to consider the path that you took when you drove to work? Perhaps you recall a time that you communicated with a friend about a very specific topic and shortly afterwards you found yourself inundated with related ads on Facebook, Gmail and YouTube. Do you ever wonder why Netflix or YouTube make the recommendations that they do? Algorithms are everywhere. You use an algorithm when finding the fastest path to work. Smart speakers, email accounts and chat programs are listening in on your conversations so they can advertise to you. Netflix and YouTube aim to maximize your viewing time and have little concern about whether or not you like the content.<\/p>\n\n\n\n

This article will examine the competing definitions of algorithms, the stakeholders who are most affected by algorithms, inequities caused by the power structures behind algorithms and what happens when algorithms go wrong. Finally, the educational aspects of algorithms will be explored. Algorithms have a significant impact on the way that research is conducted by students. Algorithms are also fundamentally changing how we consume media and directly affecting the way that we write. In order to adapt to the increasing role of algorithms in education, this article will explore creative content creation as a viable alternative to research and writing assignments that algorithms have turned into a less creative, more routinized process. Finally, this article will discuss why teachers must be aware of the influence of algorithms and work to promote algorithmic literacy among their students. <\/p>\n\n\n\n

<\/a>What are Algorithms?<\/strong><\/strong><\/h2>\n\n\n\n

At the most basic level, an algorithm is a set of rules or instructions to be followed (Tutt, 2017)<\/a>. We make use of algorithmic thinking in our everyday lives. We use an algorithm when we determine the fastest way to drive to work, when we navigate through a cafeteria in the most efficient way possible or when we follow a recipe to bake a cake (Wing, 2006)<\/a>. Algorithms are also a fundamental component of how computers function, providing a desired output based on inputs (Lloyd, 2019)<\/a>. The tasks performed by computer algorithms range from simple to complex. One example of a simple algorithm is a sort function, an algorithm that organizes entities numerically or alphabetically (V\u00f6cking et al., 2010)<\/a>. Manufacturers embed simple algorithms within the electronics of nearly all modern appliances, performing basic functions like cooking rice or moderating the temperature of the fridge (V\u00f6cking et al., 2010)<\/a>. Technologies such as GPS units, that are now commonplace in cars and mobile phones for navigation, use more advanced algorithms. The advanced algorithms in a GPS unit simultaneously consider many variables, such as the speed of the earth\u2019s rotation and precise timings, in order to make informed predictions (V\u00f6cking <\/a>et <\/a>al., 2010)<\/a>. Other famous algorithms of this type include the original PageRank algorithm used by Google to create a database of the internet and return relevant web pages. A simplified description of how PageRank functions is that it looks at the links between web pages and assigns scores based on the popularity of the originating website (Tutt, 2017)<\/a>. A page that is being linked to from the New York Times will receive a higher PageRank score than a page that is being linked to from a personal blog (Tutt, 2017)<\/a>. While the simple and advanced algorithms described above vary in complexity, they share several similarities. All the aforementioned algorithms might be called traditional algorithms. Traditional algorithms have been designed entirely by a programmer or group of programmers who have a concrete idea of the function of each line of code, and their output is deterministic, meaning that given the same set of inputs the algorithm will always produce the same result (Tutt, 2017)<\/a>.<\/strong><\/p>\n\n\n\n

The first functional use of an artificial intelligence (AI) algorithm emerged in the 1950s. Taking cues from recent advances in neurobiology, Marvin Minsky created a machine that used a randomly connected collection of electronic synapses to perform calculations (McCorduck & Cfe, 2004; Russell & Norvig, 2002)<\/a>. Many of today\u2019s most complex algorithms rely on artificial intelligence, an umbrella term which encompasses machine learning, deep learning, neural networks and numerous algorithmic implementations such as reinforcement learning and convolutional neural networks (Zubarev, 2018)<\/a>. <\/p>\n\n\n\n

AI-based algorithms are used to personalize online content for users (Giansiracusa, 2021). According to Burrell (2016)<\/a>, machine learning is typically used for classification tasks, such as organizing search results and making predictions about credit scores and the likelihood of a loan default and filtering search results. Vasily(2018) provides a simplified definition of machine learning, describing it as predicting results from incoming data with some AI thrown in. Deep reinforcement learning (deep learning) is another type of AI. Deep learning relies on neural networks, a type of algorithm which emulates how neurons in our brain function by reinforcing strong connections and trimming weak connections, and unlike traditional programming they achieve their results through weighting and repetition (Burrell, 2016)<\/a>. Within a given neural network, there are many layers of connections and, importantly, many of those layers are hidden.<\/strong> Because of the opacity that exists within a neural network, even the creator of the algorithm is unlikely to understand exactly how the algorithm reached its decision (Burrell, 2016; O\u2019Neil, 2016)<\/a>. The opacity within AI algorithms can mask biases and make it difficult or impossible to establish accountability (O\u2019Neil, 2016)<\/a>. This creates many moral and ethical ramifications that will be discussed in the following pages.<\/p>\n\n\n\n

Natural language processing (NLP) is a subset of machine learning. It uses patterns in text and linguistic conceptualization to generate text on its own by analyzing a small block of text and guessing the next word (Giansiracusa, 2021)<\/a>. Machine learning has enabled the spread of fake news,<\/p>\n\n\n\n

\u201cand other types of malicious content. With increased usage of AI and machine learning, these attacks become harder to detect and contain, according to Symantec. In 2018, the company found 592 million unique IP addresses connected to more than 1.5 billion malicious emails. Researchers from Stanford University found more than 1.5 million AI-generated fake accounts on Twitter and generated almost 100,000 tweets that generated at least 1,000 retweets in a week. While there have been some advancements in AI, it\u2019s difficult to measure its true impact on a variety of different industries and businesses. The technology that the market is investing in is advancing so quickly that it is difficult to properly evaluate.\u201d<\/em><\/p>\n\n\n\n

Many individuals believe that they could never be tricked by AI generated writing. However, the italicized text above was generated through machine learning[1]<\/sup><\/a>. The best description of language model outputs was made by Twitter user Julian Togelius (@togelius) who Tweeted \u201cGPT-3 often performs like a clever student who hasn’t done their reading trying to bullshit their way through an exam. Some well-known facts, some half-truths, and some straight lies, strung together in what first looks like a smooth narrative\u201d (Togelius, 2020)<\/a>. As NLP systems have improved, it can become increasingly challenging to notice inconsistencies in generated text and to differentiate the errors made by an AI system with errors made by a human author, although differentiating between a computer and a human becomes easier when texts are longer or specific questions are asked (Tegmark, 2017)<\/a>. <\/p>\n\n\n\n

NLP systems have been associated with the generation of \u201cfake news\u201d content.  Giansiracusa (2021)<\/a> explains that advertising revenue is tied to the number of views that a webpage receives, encouraging content creators to create blog posts and news stories that are sensationalist, exaggerated and sometimes untrue to encourage web traffic. The impact of fake news stories can be amplified by increased attention as a result of algorithms that repost content and aggregate news stories (Giansiracusa, 2021)<\/a>.    <\/p>\n\n\n\n

In 1950, Alan Turing devised \u201cthe imitation game\u201d a test which is now known as the \u201cTuring Test\u201d, to distinguish between humans and machines based upon a natural language conversation between a computer participant (A) a human participant (B) with  a human observer (C) (Turing, 1950)<\/a>. MIT physicist and machine learning researcher Max Tegmark believes that the most advanced NLP systems are taking steps towards passing this test, although he cautions that the test is \u201ctesting human gullibility more than true artificial intelligence\u201d (Tegmark, 2017, p. 119)<\/a>. While computers and their associated algorithms have not reached a state that might be considered \u201cintelligence\u201d, the social power that algorithms possess can influence human agency (Lloyd, 2019)<\/a> and the effectiveness of algorithms in creating deceptive content such as fake news will require us to re-examine our critical literacy practices.<\/p>\n\n\n\n

<\/a>Interpreting Algorithms<\/strong><\/strong><\/h2>\n\n\n\n

We interpret and interact with algorithms in different ways, based upon our individual awareness of algorithms, our personal technical expertise, whether we have access to the algorithm’s code, and the complexity of an algorithm’s underlying code. Algorithmic culture has become a larger part of our everyday lives and it influences the choices and decisions that we make on an everyday basis (Lloyd, 2019)<\/a>. For those not immersed in the world of algorithms, the results of algorithms often appear to be magic or go unnoticed. This type of user is unlikely to notice subtle changes in Amazon recommendations based on search histories or the influence that past searches have made on their Google search results. They are unaware of the datafication of our lives, a process where our everyday practices are measured and recorded, allowing for data analysis, including tracking and prediction (Mertala, 2020)<\/a>. A private conversation about cars might be overheard by a smart speaker and lead to an increased quantity of automobile advertisements. Lloyd (2019)<\/a>, notes that this group of users knowingly or unknowingly give up authority to algorithms when they perform searches or interact with technology. Assuming that this group of na\u00efve computer users is composed of Baby Boomers and GenXers is a common assumption, non-digital-natives who grew up without exposure to advanced technologies and the powerful influence of algorithms should be less likely to possess high levels of digital literacy (Henderson et al., 2020)<\/a>. However, Henderson at al. (2020)<\/a> <\/strong>has found that age is a worse predictor of digital literacy than gender, ability, and socioeconomic status. If simple exposure to technology and algorithms does not increase literacy, what can be done to educate today\u2019s generation of digital natives?<\/p>\n\n\n\n

For those with a deeper understanding of algorithmic interactions, two definitions emerge. The first definition is given by the individuals who create algorithms. For mathematicians, programmers, engineers and alike, algorithms are viewed as a computer function that provides a desired output based on a series of inputs (Lloyd, 2019)<\/a>. <\/em><\/strong>Algorithms are viewed less positively by media theorists, sociologists and others who study algorithms from an outsider’s perspective who believe that algorithms contain inherent cultural biases and that they lack accountability and transparency for the decisions they make whose impact can range from mundane to life-changing (Lloyd, 2019; O\u2019Neil,<\/a> 201<\/a>6)<\/a>.<\/p>\n\n\n\n

<\/a>Algorithms from an Insider\u2019s perspective<\/strong><\/strong><\/h2>\n\n\n\n

For computer scientists, mathematicians, and engineers; algorithms are often described as a component of computational thinking. Computational thinking focuses on the mathematical process of abstraction and includes algorithms, data structures, state machines, programming languages, logistics and semantics, heuristics, control structures, communication and architectures (Wing, 2006)<\/a> Computer scientists, engineers and other creators of algorithms interact with algorithms from the inside. For the creators of algorithms, algorithms are often viewed with extreme positivity. V\u00f6cking et al. describe algorithms as \u201cbeautiful ideas for tackling or solving computational problems more efficiently\u201d (2011, p. vi)<\/a>. Computer scientists and engineers rightfully praise algorithms for their ability to solve complex problems and the many scientific achievements that would not be possible without AI and advanced algorithms, including the mapping of the human genome, prediction of weather events and the development of medicines and vaccines (V\u00f6cking et al., 2010; Wing, 2006)<\/a>. Algorithms and data collection are advertised as a way to improve the education sector by Holmes et al. (2016)<\/a> who argue that Educational Data Mining (EDM) can be used to track students’ behaviours and identify students that are at risk of dropping out Additionally, AI can used to provide personalized tutoring to students and facilitate group work by identifying students that would work well as a team based on cognitive abilities and interests (Holmes et al., 2016)<\/a>.<\/p>\n\n\n\n

While tech industry insiders are optimistic about the power of algorithms, they are also acutely aware of the potential negative impacts of technology. Steve Jobs famously restricted his children from using iPads (Bilton, 2014)<\/a> and an increasing number of Silicon Valley tech employees are banning children\u2019s use of technology at home and sending their children in tech-free schools (Weller, 2018)<\/a>. The promotion of technology by powerful tech figures while simultaneously limiting their families exposure to technology, indicates that technology use might also have negative impacts.<\/p>\n\n\n\n

<\/a>Algorithms from the outside<\/strong><\/strong><\/h2>\n\n\n\n

In the movie Jurassic Park, Jeff Goldblum\u2019s character states \u201cyour scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.\u201d (Spielberg, 1993)<\/a> Media theorists, sociologists and proponents of social justice have an opposing view of what data is and they see a darker side to algorithms. Many in this group view our data as an extension of ourselves. Cheney-Lippold (2017) explains that humans exist within the digital realm as digital entities made up of bits and our digital entities are categorized by algorithms. Tegmark (2017)<\/a> extends the view that humans are data even further, envisioning a future in which data would enable humans to exist after death. The framing of humans as data entails that data needs to be handled responsibly in the same way that people are; however, this view is incompatible with large tech companies who view data as a commodity (Raffaghelli, 2020)<\/a>. Categorizations made by Google are often wrong, with Cheney-Lippold (2017) providing the example that women who use Google to conduct research on topics within male-dominated fields are more likely to be gendered as male by Google\u2019s algorithm.<\/p>\n\n\n\n

Cathy O\u2019Neil coined the term \u201cweapons of math destruction\u201d to refer to a group of algorithms that are opaque, large scale and damaging (2016)<\/a>. O\u2019Neil\u2019s criticisms of algorithms are shared by other theorists who are concerned with the reach of algorithms, a lack of accountability, and the destruction that algorithms have caused and might cause in the future; topics that will be explored within the following sections.<\/p>\n\n\n\n

In addition to O\u2019Neil\u2019s specific criticisms, there is also a general concern about the growth in usage of AI-powered bots to spread election propaganda and other types of fake news as well as the possibility that large companies who are in possession of large amounts of human data could make use of their large data sets and AI to manipulate behaviour.(2020)<\/a><\/p>\n\n\n\n

The use of AI within education is a contentious issue. Mertala (2020)<\/a> is concerned with the datafication of education proposed by Holmes, arguing that there are negative social consequences to the collection of attendance and test scores, which are often used as a way to compare schools and measure student and teacher performance. Holmes et al.\u2019s (2016) support of EDM to identify at-risk students is also problematic for Raffaghelli et al. (2020)<\/a> who are critical of the machine learning systems used to identify vulnerable students, noting that the labelling stays with the student throughout their studies and might cause a student to become more vulnerable. Raffaghelli et al. (2020) go on to suggest that any analysis of vulnerability should be considered within the context of student agency and that data practices need to be transparent and developed with the participation of staff and students.<\/p>\n\n\n\n

<\/a>Algorithms are opaque<\/strong><\/strong><\/h2>\n\n\n\n

AI algorithms are extraordinarily difficult for external researchers to understand and analyze for several reasons. Algorithms falling under the umbrella of AI have been described as a \u201cblack box\u201d because, in most cases, the source code for the algorithm is proprietary and not available to the public (Burrell, 2016; Lloyd, 2019)<\/a>. Burrell (2016) breaks down the opaque nature of algorithms into three distinct categories: intentional opacity, opacity due to technical illiteracy, and opacity due to scale. Intentional opacity often occurs because companies view their algorithms as intellectual property and closely guard their algorithms from any type of external inspection (O\u2019Neil, 2016)<\/a>.  The second category of algorithmic opacity is technical illiteracy, with Burrel (2016) noting that reading and writing code and understanding the design of algorithms is a specialized skill that the majority of the population does not possess. However, even if the design of an algorithm is understood it is unlikely that a human programmer could determine how a specific result was achieved as \u201cboth the inputs (data about humans) and outputs (classifications) can be unknown and unknowable\u201d (Mittelstadt et al., 2016, p. 6)<\/a>.<\/strong> Examples of opacity due to scale include the algorithms that underlie Google\u2019s search functionality which are so large and complex that teams of engineers each have an understanding of a small component of the search function and no single person understands the algorithm in its entirety Because algorithms are opaque, there is no mechanism to audit an algorithm\u2019s findings, if the results generated by the algorithm are unfair or discriminatory  (Mittelstadt et al., 2016; O\u2019Neil, 2016).The opacity of algorithms makes the results of algorithms difficult to challenge and reduces the agency of individuals who the algorithm is affecting (Mittelstadt et al., 2016)<\/a><\/p>\n\n\n\n

<\/a>Discriminatory Algorithms and the Power Structures that Drive them<\/strong><\/strong><\/h2>\n\n\n\n

The incongruent views of algorithms between scientists and theorists could be related to the power differential that exists between the owners of algorithms and the users of algorithms. <\/p>\n\n\n\n

According to Lloyd (2019)<\/a>, power structures have an influence on the creation of algorithms themselves. Many of the most advanced and influential algorithms in use today are developed by a group of companies coined \u201cFAANG\u201d composed of Facebook (now Meta)<\/em>, Amazon, Apple , Netflix, and Google (Alphabet)<\/em> (Hobbs, 2020)<\/a>. All the companies within FAANG are based in The United States of America but have a global influence. (Lloyd, 2019)<\/a>The programmers, engineers and computer scientists that work for the FAANG group of companies are predominantly high socio-economic status males and when computer code is generated, it is influenced by the social power and positionality of the coder (Lloyd, 2019)<\/a>. In addition to the biases held by the programmers themselves, AI-based algorithms generally require training data, consisting of inputs that can be used to teach the algorithm the compilation of datasets used to train AI algorithms, and these datasets are not neutral (L<\/a>ean<\/a>der & Burriss, 2020)<\/a>. When facial detection systems were first created, computer scientists used their own faces to train the algorithms and, as a result, facial detection systems are most effective at identifying white male faces (O\u2019Neil, 2016)<\/a>. Modern implementations of facial detection algorithms have used datasets such as the faces of US military employees, imparting a gender bias into the functioning of AI systems (Leander & Burriss, 2020)<\/a>. While algorithms have been shown to give preferential treatment to individuals that resemble their creators, they have also been found to discriminate against outsiders. A notable and particularly egregious example is Level of Service Inventory\u2013Revised (LSI-R), a statistics-based sentencing model that relies on a questionnaire and recidivism data to help determine sentencing in at least twenty-four US States (O\u2019Neil, 2016)<\/a>. According to O\u2019Neil (2016) the questionnaire contains questions that are inadmissible in court and the LSI-R model consistently recommends harsher sentencing for African Americans even when the crimes are similar. <\/p>\n\n\n\n

<\/a>The “Whoops” Factor – Algorithms that have Unintended Influence\/Impact<\/strong><\/h3>\n\n\n\n

As algorithms become more and more complex, they do not always function in ways their creators intended. Microsoft\u2019s \u201cTAY\u201d an acronym for “thinking about you” was an online chatbot released by Microsoft in 2016 (Lee, 2016)<\/a>. According to Microsoft (20<\/a>1<\/a>6)<\/a>, TAY was created by \u201cmining relevant public data and by using AI and editorial developed by a staff including improvisational comedians\u201d and was designed to interact with 18 – 24-year-olds in the US. In less than a day, TAY transformed from a friendly chat bot with a teenage persona to vehement Trump supporter that generated racist, pro-nazi and anti-feminist tweets (Hunt, 2016)<\/a>. The TAY experiment lasted for less than 24 hours before the bot was taken down and Microsoft provided an official apology (Microsoft, 2016). Malfunctioning algorithms can also lead to disastrous effects, including the loss of life. An example of algorithms leading to loss of life is illustrated by the recent Boeing 737 max crashes. The crashes were caused by several factors, first a malfunctioning angle of attack indicator (AOA) sent false data to the airplane’s flight computer. The incorrect AOA data led algorithms within the flight computer\u2019s Maneuvering Characteristics Augmentation System (MCAS) to make adjustments to the flight control surfaces, causing the plane\u2019s nose to pitch down, leading to the Lion Air Flight 610 crash on October 29, 2018, and Ethiopian Airlines Flight 302 crash on March 10, 2019, resulting in the deaths of 346 people (Gelles, 2019; Perell, 2020). Computer algorithms may have caused Wall Street\u2019s trillion-dollar \u201cFlash Crash” of May 6, 2010 (Tegmark, 2017). The Flash crash of May 6, 2010 caused the US stock indices to fall by nearly 9% within 15 minutes leading to the loss of over a trillion dollars of market value (Lin, 2013). High frequency trading (HFT) which uses algorithms to make stock trades more quickly than any human broker are a primary cause of the crash (Lauricela et al., 2010). Also implicated, was a single, home-based, trader who manipulated the market by using a computer program to create fake sell orders. (Lynch & Meidema, 2015). Tegmark (2017) believes that one of the root causes of accidents is that algorithms do not understand the data that they are interacting causing them to make predictions that a human would immediately recognize as invalid and he argues that all algorithms require both validation and verification.<\/p>\n\n\n\n

Another unintended consequence of algorithms is that algorithms exacerbate the disadvantages faced by marginalized populations who may be denied access to loan applications, college admissions and employment based on their zip code, which is often a proxy for race and socioeconomic status (O\u2019Neil, 2016)<\/a>. Mittelstadt et al. (2016)<\/a> define this type of algorithmic discrimination as \u201cunfair outcomes\u201d, which occur when algorithms make data-driven conclusions that are discriminatory. As discussed within the \u201cAlgorithms are Opaque\u201d section, it can be difficult to prove that algorithms are acting in a discriminatory manner. To reduce the harm caused by discriminatory algorithms, O\u2019Neil (2016) recommends that algorithms be tested with individuals from diverse backgrounds to determine if the algorithms are biased and believes that government regulation is also required.<\/p>\n\n\n\n

Algorithms and Education<\/strong><\/strong><\/h2>\n\n\n\n

How Technology and Algorithms are Transforming Education<\/strong><\/strong><\/h2>\n\n\n\n

For as long as computers have existed, education technologists have made claims that computers, the internet and student access to new technologies will transform education and create equitable educational outcomes (Reich, 2020)<\/a>. However, increased access to technology, such as computers, has not been found to improve educational outcomes (Waluyo, 2019; Starkey & Zhong, 2019). Battle (1999) finds that students from higher socioeconomic backgrounds are more likely to use computers for research purposes and that they receive a greater educational benefit from computer use. There is a real threat that the increased use of technology will increase the digital divide which the OECD (2001) describes as \u201cthe gap between individuals, households, businesses, and geographic areas at different socioeconomic levels concerning both their opportunities to access information and communication technologies and their use of the Internet for a wide variety of activities\u201d (p.5). Better teacher training has been found to improve the benefits that students receive from the use of technology reducing the digital divide (Starkley et al., 2016; Warcheauer et al., 2016). Reich (2020) recommends that educators immerse themselves in new technologies and teach students how to use new technologies, rather than continuing to teach in old ways and using technology as a tool. A similar strategy is recommended for increasing artificial intelligence literacy among students. To increase artificial intelligence literacy, Reich (2020) advocates that students learn about how AI functions and ethical considerations when working with AI rather than just learning how to use AI-based programs (Ng et al., 2021). In order to prepare students for a world filled with algorithms and AI educators will first need to understand these technologies so that they can provide effective artificial intelligence literacy instruction.<\/p>\n\n\n\n

<\/a>How Algorithms shape the media that we consume<\/strong><\/strong><\/h2>\n\n\n\n

As storage and processing power have become affordable, there is no limit to the amount of information that can be stored and analyzed (Cheney-Lippold, 2017)<\/a>. Algorithms make use of this large body of information to assign age, gender and race to users based on their search history and identified interests (Cheney-Lippold, 2017)<\/a>. Our internet histories can have a direct influence on what we read. When we perform a google search to find more information about a topic or for suggestions for the next book to read, Google considers our stored personal data and search keywords before returning its results (Kraus-Friedberg, 2021)<\/a>. The results that we see are also reduced by our quest for efficient retrieval of information. Bhatt and MacKrenzie (2019)<\/a> found that higher education students in the fields of STEM, the humanities and business all used ritualized behaviours in their searches related to course assignments. A broad analysis of computer users found that during a Google search, somewhere between 30% and 40% of all users click on the first result, and two-thirds of users click on one of the first three results (Bailyn, 2022; Fay, 2022)<\/a>. Google\u2019s search algorithm has made it easy to find information. Unfortunately, the same filtering that makes Google searches so efficient also works to filter out alternate perspectives that might be less popular or that run contrary to the profile that Google has established for us based on our previous search history. While the specific details of the PageRank algorithm is a closely guarded secret, Google has released the details of how its YouTube Recommendation system works. On YouTube, the effects of prior searches are especially apparent because Google\u2019s recommendation algorithm has been finely tuned to maximize the amount of time that users spend consuming video content (Giansiracusa, 2021)<\/a>.  Giansiracusa\u2019s recommendation algorithm findings are mirrored in Google\u2019s own research presentation. At the time of Google\u2019s 2016 conference presentation, the algorithm for video recommendations was described as \u201cgenerally a simple function of expected watch time per impression.\u201d that \u201cis constantly being tuned based on live A\/B testing results\u201d (Covington et al., 2016, p. 5)<\/a> Figure 3 of Google\u2019s presentation Deep Neural Networks for YouTube Recommendations. Proceedings of the 10th ACM Conference on Recommender Systems, 191\u2013198<\/em> indicates that a user\u2019s viewing history has an impact on the type of videos that are recommended and the ranking of the recommended videos. To counter the bias that exists within internet research, Bhatt and MacKenzie (2019)<\/a> believe that students need to learn about algorithms to better understand the way that their online experiences are shaped and to make students aware that any internet research contains bias.<\/p>\n\n\n\n

<\/a>How Algorithms are Transforming Writing<\/strong><\/strong><\/h2>\n\n\n\n

Algorithms have transformed societal writing patterns. The widespread use of mobile phones has replaced simple writing tasks, such as to-do lists, grocery lists and directions. While early mobile phones required users to provide input through typing, improvements in the machine learning algorithms that power voice to text translation have increased the likelihood that information is dictated to our devices rather than typed. According to a Stanford study, speech recognition using Baidu\u2019s Deep Speech 2 deep learning-based algorithm was three times faster than typing on an iPhone and resulted in fewer errors (Ruan et al., 2017)<\/a>. According to the Ruan et al. (2017) study, speech recognition could transcribe text at a rate of 100 words per minute, a rate that is significantly faster than students in the study could handwrite or type on a computer keyboard (Horne et al., 2011)<\/a>.<\/p>\n\n\n\n

Technologies designed to improve a user\u2019s writing, such as Grammarly’ and ProWritingAid’, have also made improvements in their ability to find and correct errors in writing. An action research study from Indonesia found that the use of Grammarly improved essay writing, increasing the number of students meeting the success criteria from 62% to 80% with key improvements in the categories of grammar and diction (Karyu<\/a>a<\/a>try, 2018)<\/a>. It is conceivable that programs designed to correct spelling and grammar errors could become so proficient that they would find and correct virtually every error within a given text. When programs reach this critical threshold, the impact could be as great as the impact of the calculator on mathematics. Although spelling and grammar knowledge would still be important, assessment of written works would have a greater focus on creativity and originality and a reduced focus on spelling, punctuation and syntax. <\/p>\n\n\n\n

Tools such as Grammarly and ProWritingAid correct what you type, the newest generation of tools do the typing for you. Based on natural language processing and deep learning AI models, autocomplete technologies provide word suggestions as you type (Tegmark,<\/a> <\/a>2017)<\/a>. One example of an autocomplete technology is Google\u2019s Smart Compose, introduced to Gmail around 2009 as a way to make typing emails faster, it has since been added to other Google applications including the Google Docs application used in many schools (Google, 2022). According to information presented by Google at the 2019 Knowledge, Discovery and Datamining Conference, \u201cthe core of Smart Compose is a large-scale neural language model\u201d (Chen et al., 2019, p. 2287). Smart Compose makes use of large language models that have been trained using machine learning, by looking at millions of sentences that have been written by others, the algorithm predicts the words that are most likely to come next in your sentence and offers them as suggestions that can be automatically added to your writing by pressing the tab key (Chen et al., 2019). Technologies such as Smart Compose are problematic for educators because the suggestions provided reflect the most common words written by adults. When students accept the recommendations made by Smart Compose it may change the flow and style of a student\u2019s writing. More significantly, text completion algorithms have the potential to impart an unintended bias in writing. Google (2022) acknowledges the possibility of unintended bias by posting the following disclaimer on its smart compose web page: As language understanding models use billions of common phrases and sentences to automatically learn about the world, they can also reflect human cognitive biases. Being aware of this is a good start, and the conversation around how to handle it is ongoing. Google is committed to making products that work well for everyone and are actively researching unintended bias and mitigation strategies (2022).<\/p>\n\n\n\n

In the coming years, educators will need to cope with writing that has been generated entirely by algorithms. In 2005, Massachusetts Institute of Technology (MIT) students created a rudimentary program called \u201cSCIgen\u201d that generated nonsensical research papers. The students submitted the fake papers and were invited to speak at the World Multiconference on Systemics, Cybernetics and Informatics (WMSCI) to present their entirely generated paper \u201cRooter: A Methodology for the Typical Unification of Access Points and Redundancy,\u201d (Conner-Simons, 2015)<\/a>. Springer Publishing, one of the largest academic publishing houses, announced that they had removed over 120 papers generated by SCIgen, and had created a free tool to detect papers created by SCIgen (Conner<\/a>–<\/a>Simons, 2015)<\/a>. Modern algorithms relying on advanced language models produce text that is much more convincing than the text generated by SCIgen. Given an input of several words, the natural language processing algorithms (such as the aforementioned GPT-J) are able to generate text of any length at speeds that are significantly faster than any human can generate content. An example recalled by Giansiracusa (2021)<\/a>, was a GPT-3 bot that posed as a human on Reddit\u2019s \/r\/AskReddit forum and responded to questions in under one minute with six paragraph long responses. As research and the generation of text becomes simplified, I predict that there will be a change in the way that knowledge is produced and presented within education, with research being aided by algorithms and the production of text becoming more efficient. Bhatt and MacKenize (2019) advocate for a shift away from ritualized student research towards the creation of original content. A view is compatible with the prediction that any job that can be performed on a computer will eventually be taken over by computers (Reich, 2020; Tegmark, 2017)<\/a><\/p>\n\n\n\n

<\/a>The need for Data Literacy<\/strong><\/strong><\/h2>\n\n\n\n

Henderson et al. (2020) challenged the concept of \u201cdigital, or the idea that those who have been brought up during a time of advanced technology are automatically experts in technology use . In their study, young adults (19-29) were found not to demonstrate advanced digital skills and assumptions of technological expertise and digital fluency was damaging to educational development. The rapid growth in the way that AI driven systems have shaped the internet has not been reflected in \u201ccurrent educational curricula, practices and theories\u201d (Leander & Burriss, 2020, p. 1264). An example of this is the concept of a shared text, texts are no longer static, they can be customized to the individual and may change over time we can no longer assume that texts are a shared experience (Leander & Burriss, 2020). According to Henderson et al (2020), engaging students in internet research is transformative and can have a lasting impact on students\u2019 digital lives beyond school. Mertala (2020) has found that the education sector has enacted formal data literacy pedagogies that focus on improving students’ data literacy when conducting research, but that they neglect to teach students about data literacy in their everyday lives. Henderson et al. (2020) advocate the use of a digital literacy toolkit that includes the development of skills in a participatory manner and encourages students to investigate \u201cdataveillance , internet wellbeing, and artificial intelligence\/machine learning\u201d as well as consider personal privacy issues. Before such a toolkit could be implemented, frameworks are needed to guide educators in teaching students about AI (Ng et al., 2021).<\/p>\n\n\n\n

Traditional approaches to literacy and the analysis of media will need to be revised in response to the growing impact of algorithms. Although media educators have pedagogical tools to critically analyze texts, Leander and Burriss (2020) express concern that new media is often entangled with machines and, for this reason, the methods used to analyze text need to expand. The definition of the hidden curriculum,  lessons which are taught and learned in an unconscious manner (Kentli, 2009) needs to be broadened to include data literacy, as the materials used within a classroom, such as computers and software, might also teach students about hidden messages (Mertala, 2020). The definition of literacies also needs to be expanded to include computer coding. According to Vee (2017), coding parallels literacy as it is \u201ca socially situated, symbolic system that enables new kinds of expressions as well as the scaling up of pre-existing forms of communication\u201d (p. 4). Within a new definition of literacy that includes computer coding, critical computer literacy should also be explored. While understanding technology is an excellent first step, users of technology must also recognize the impact of AI and take action towards AI that is unjust (Leander & Burriss, 2020). As teachers we need to promote data literacy among our students. According to Raffaghelli (2020) Strong data literacy skills enables students to express themselves, it enhances diversity, and helps students contribute towards social justice issues.<\/p>\n\n\n\n

<\/a>Algorithms and the Future Direction of Education<\/strong><\/strong><\/h2>\n\n\n\n

Tegmark (2017) predicts that in the future of employment, the skills that are least likely to be replaced by AI include social intelligence, creativity, and working in an unpredictable environment. Reich (2020) calls for educational reforms, noting that any metric that can be easily accessed by a computer is likely related to a task that can be easily performed by a computer and not required within the labour market with examples including bank tellers and airline check-in employees. One potential way for students to express creativity and social intelligence is through content creation through platforms such as TikTok, a video sharing platform that lets users record brief clips that can then be edited to include filters and background music. While TikTok has a reputation among parents of providing little educational value and questionable content (Quirt, 2020) the videos created on the platform show a high level of creativity and resonate with today\u2019s youth in a way that Anderson (2020) describes as \u201ccreative chaos\u201d due to the wide variety of content and the uncertainty of what will appear next (p. 5). Using a thematic analysis approach, Li et al. (2021) found that TikTok videos and, in particular, TikTok dances were one of the most effective ways to freely convey public health information related to Covid-19 based upon the number of likes and shares. TikTok can also be used to gauge student sentiment. Using a thematic analysis approach, Literat (2021) studied the content of 1,930 TikTok videos containing the hashtag #onlineschool finding that students attending school virtually had significant concerns about mental health issues and the way that their home life would be perceived by others. In its current state, TikTok is not a viable solution for education, as some videos contain explicit content and there are also concerns about online privacy and the algorithms used by the platform to determine which videos will be displayed next (Anderson, 2020). However, there are also many unexpected positives to TikTok, with Literat (2021) noting that students created videos to help others with math concepts and that the TikTok comments section was used for peer support and \u201cmental health check-ins on videos tagged #ADHD\u201d (p. 7). As educators, we need to be aware of the wide sweeping changes that algorithms have made to the education system and embrace new methods of communication in order to resonate with students.<\/p>\n\n\n\n

Conclusion<\/strong><\/h2>\n\n\n\n

Algorithms will continue to become an increasingly prevalent part of our daily lives. While some scholars have called for the regulation of algorithms (O\u2019Neil, 2016; Tutt, 2017)<\/a> the global spread of the internet makes regulation difficult to implement on a global scale. As computers become faster and algorithms become more advanced, I believe that the impact of algorithms and artificial intelligence will continue to grow. Algorithms have the ability to make our lives easier by simplifying communication. Algorithms can increase our media enjoyment by providing us content that is curated to meet our interests. But algorithms have also been shown to reflect the worst human qualities, including callousness when making life impacting decisions, racism, and the pursuit of profit above all else. The impacts of algorithms on education are widespread, with algorithms affecting how we receive, process, and communicate information. Educators will need to adapt to the changing employment climate brought on by algorithms and also teach students the algorithmic literacy skills that are necessary to understand the influence of algorithms. <\/p>\n\n\n\n

References<\/strong><\/h2>\n\n\n\n

Anderson, K. E. (2020). Getting acquainted with social networks and apps: It is time to talk about TikTok. Library Hi Tech News<\/em>.<\/a><\/p>\n\n\n\n

Bailyn, E. (2022, January 5). Google Click-Through Rates (CTRs) by Ranking Position in 2022. First Page Sage<\/em>. https:\/\/firstpagesage.com\/seo-blog\/google-click-through-rates-ctrs-by-ranking-position-in-2022\/<\/a><\/p>\n\n\n\n

Battle, P. A., Juan. (1999). Home Computers and School Performance. The Information Society<\/em>, 15<\/em>(1), 1\u201310. https:\/\/doi.org\/10.1080\/019722499128628<\/a><\/p>\n\n\n\n

Bhatt, I., & MacKenzie, A. (2019). Just Google it! Digital literacy and the epistemology of ignorance. <\/a>Teaching in Higher Education<\/em><\/a>, <\/a>24<\/em><\/a>(3), 302\u2013317.<\/a><\/p>\n\n\n\n

Bilton, N. (2014, September 10). Steve Jobs Was a Low-Tech Parent. <\/a>The New York Times<\/em><\/a>. https:\/\/www.nytimes.com\/2014\/09\/11\/fashion\/steve-jobs-apple-was-a-low-tech-parent.html<\/a><\/p>\n\n\n\n

Burrell, J. (2016). How the machine \u2018thinks\u2019: Understanding opacity in machine learning algorithms. <\/a>Big Data & Society<\/em><\/a>, <\/a>3<\/em><\/a>(1), 205395171562251. https:\/\/doi.org\/10.1177\/2053951715622512<\/a><\/p>\n\n\n\n

Cheney-Lippold, J. (2017). <\/a>We are data: Algorithms and the making of our digital selves<\/em><\/a>. New York University Press.<\/a><\/p>\n\n\n\n

Conner-Simons, A. (2015). <\/a>How three MIT students fooled the world of scientific journals<\/em><\/a>. MIT News | Massachusetts Institute of Technology. https:\/\/news.mit.edu\/2015\/how-three-mit-students-fooled-scientific-journals-0414<\/a><\/p>\n\n\n\n

Covington, P., Adams, J., & Sargin, E. (2016). Deep Neural Networks for YouTube Recommendations. <\/a>Proceedings of the 10th ACM Conference on Recommender Systems<\/em>, 191\u2013198. https:\/\/doi.org\/10.1145\/2959100.2959190<\/a><\/p>\n\n\n\n

Fay, O. (2022, March 31). The Value Of Google Result Positioning. <\/a>Poll the People<\/em><\/a>. https:\/\/pollthepeople.app\/the-value-of-google-result-positioning-3\/<\/a><\/p>\n\n\n\n

Foucault, M. (1980). Power\/knowledge: Selected interviews and other writings, 1972-1977<\/em>. Vintage.<\/p>\n\n\n\n

Gelles, D. (2019, March 22). Boeing 737 Max: What\u2019s Happened After the 2 Deadly Crashes. <\/a>The New York Times<\/em><\/a>. https:\/\/www.nytimes.com\/interactive\/2019\/business\/boeing-737-crashes.html, https:\/\/www.nytimes.com\/interactive\/2019\/business\/boeing-737-crashes.html<\/a><\/p>\n\n\n\n

Giansiracusa, N. (2021). <\/a>How Algorithms Create and Prevent Fake News: Exploring the Impacts of Social Media, Deepfakes, GPT-3, and More<\/em><\/a>. Apress. https:\/\/doi.org\/10.1007\/978-1-4842-7155-1<\/a><\/p>\n\n\n\n

Henderson, M. J., Shade, L. R., & Mackinnon, K. (2020). EVERY CLICK YOU MAKE: ALGORITHMIC LITERACY AND THE DIGITAL LIVES OF YOUNG ADULTS. <\/a>AoIR Selected Papers of Internet Research<\/em><\/a>.<\/a><\/p>\n\n\n\n

Hobbs, B. (2020, November 6). <\/a>FAANG stands for five very successful tech companies that can move the stock market\u2014Here\u2019s what to know about investing in them<\/em><\/a>. Business Insider. https:\/\/www.businessinsider.com\/what-is-faang<\/a><\/p>\n\n\n\n

Holmes, W., Luckin, R., & UCL Knowledge Lab. (2016). <\/a>Intelligence unleashed: An argument for AI in education<\/em><\/a>. Pearson\u202f: UCL Knowledge Lab.<\/a><\/p>\n\n\n\n

Horne, J., Ferrier, J., Singleton, C., & Read, C. (2011). <\/a>Computerised assessment of handwriting and typing speed<\/em><\/a>. <\/a>28<\/em><\/a>(2), 16.<\/a><\/p>\n\n\n\n

Hunt, E. (2016, March 24). Tay, Microsoft\u2019s AI chatbot, gets a crash course in racism from Twitter. <\/a>The Guardian<\/em><\/a>. https:\/\/www.theguardian.com\/technology\/2016\/mar\/24\/tay-microsofts-ai-chatbot-gets-a-crash-course-in-racism-from-twitter<\/a><\/p>\n\n\n\n

Jeppesen,S., Hoechsmann, M., ulthiin, i., VanDyke, D., McKee, M. (2022). The Capitol Riots: Digital Media, Disinformation, and Democracy Under Attack<\/em>. Routledge.<\/p>\n\n\n\n

Karyuatry, L. (2018). Grammarly as a Tool to Improve Students\u2019 Writing Quality: Free Online-Proofreader across the Boundaries. <\/a>JSSH (Jurnal Sains Sosial Dan Humaniora)<\/em><\/a>, <\/a>2<\/em><\/a>(1), 83. https:\/\/doi.org\/10.30595\/jssh.v2i1.2297<\/a><\/p>\n\n\n\n

Kraus-Friedberg, C. (2021). I Could Just Google It: Algorithms, Information Literacy and Database Instruction. <\/a>Michigan Academician<\/em><\/a>, <\/a>47<\/em><\/a>(3), 74. ABI\/INFORM Global.<\/a><\/p>\n\n\n\n

Leander, K. M., & Burriss, S. K. (2020). Critical literacy for a posthuman world: When people read, and become, with machines. <\/a>British Journal of Educational Technology<\/em><\/a>, <\/a>51<\/em><\/a>(4), 1262\u20131276.<\/a><\/p>\n\n\n\n

Lee, P. (2016, March 25). <\/a>Learning from Tay\u2019s introduction<\/em><\/a>. The Official Microsoft Blog. https:\/\/blogs.microsoft.com\/blog\/2016\/03\/25\/learning-tays-introduction\/<\/a><\/p>\n\n\n\n

Li, Y., Guan, M., Hammond, P., & Berrey, L. E. (2021). Communicating COVID-19 information on TikTok: A content analysis of TikTok videos from official accounts featured in the COVID-19 information hub. <\/a>Health Education Research<\/em><\/a>.<\/a><\/p>\n\n\n\n

Literat, I. (2021). \u201cTeachers Act Like We\u2019re Robots\u201d: TikTok as a Window Into Youth Experiences of Online Learning During COVID-19. <\/a>AERA Open<\/em><\/a><\/p>\n\n\n\n

, <\/a>7<\/em><\/a>, 2332858421995537.<\/a><\/p>\n\n\n\n

Lloyd, A. (2019). Chasing Frankenstein\u2019s monster: Information literacy in the black box society. <\/a>Journal of Documentation<\/em><\/a>, <\/a>75<\/em><\/a>(6), 1475\u20131485. https:\/\/doi.org\/10.1108\/JD-02-2019-0035<\/a><\/p>\n\n\n\n

McCorduck, P., & Cfe, C. (2004). <\/a>Machines who think: A personal inquiry into the history and prospects of artificial intelligence<\/em><\/a>. CRC Press.<\/a><\/p>\n\n\n\n

Mertala, P. (2020). Data (il) literacy education as a hidden curriculum of the datafication of education. <\/a>Journal of Media Literacy Education<\/em><\/a>, <\/a>12<\/em><\/a>(3), 30\u201342.<\/a><\/p>\n\n\n\n

Microsoft. (2016, April 14). <\/a>Meet Tay\u2014Microsoft A.I. chatbot with zero chill<\/em><\/a>. https:\/\/web.archive.org\/web\/20160414074049\/https:\/\/www.tay.ai\/<\/a><\/p>\n\n\n\n

Mittelstadt, B. D., Allo, P., Taddeo, M., Wachter, S., & Floridi, L. (2016). The ethics of algorithms: Mapping the debate. <\/a>Big Data & Society<\/em><\/a>, <\/a>3<\/em><\/a>(2), 205395171667967. https:\/\/doi.org\/10.1177\/2053951716679679<\/a><\/p>\n\n\n\n

Ng, D. T. K., Leung, J. K. L., Chu, S. K. W., & Qiao, M. S. (2021). Conceptualizing AI literacy: An exploratory review. <\/a>Computers and Education: Artificial Intelligence<\/em><\/a>, <\/a>2<\/em><\/a>, 100041. https:\/\/doi.org\/10.1016\/j.caeai.2021.100041<\/a><\/p>\n\n\n\n

OECD Publications. 2001. Understanding the digital divide.<\/p>\n\n\n\n

http:\/\/www.oecd.org\/decument\/51\/0,3343,en_2649_33757_1814131_1_1 _1_,00.html<\/p>\n\n\n\n

O\u2019Neil, C. (2016). <\/a>Weapons of math destruction: How big data increases inequality and threatens democracy<\/em><\/a>. Broadway Books.<\/a><\/p>\n\n\n\n

Perell, D. (2020). <\/a>Why Did The Boeing 737 Max Crash?<\/em><\/a> David Perell. https:\/\/perell.com\/essay\/boeing-737-max\/<\/a><\/p>\n\n\n\n

Raffaghelli, E. (2020). Is data literacy a catalyst of social justice? A response from nine data literacy initiatives in higher education. <\/a>Education Sciences<\/em><\/a>, <\/a>10<\/em><\/a>(9), 233.<\/a><\/p>\n\n\n\n

Raffaghelli, J. E., Manca, S., Stewart, B., Prinsloo, P., & Sangr\u00e0, A. (2020). <\/a>Supporting the development of critical data literacies in higher education: Building blocks for fair data cultures in society<\/em><\/a>. Springer.<\/a><\/p>\n\n\n\n

Reich, J. (2020). <\/a>Failure to disrupt: Why technology alone can\u2019t transform education<\/em><\/a>. Harvard University Press.<\/a><\/p>\n\n\n\n

Ruan, S., Wobbrock, J. O., Liou, K., Ng, A., & Landay, J. A. (2017). <\/a>Comparing Speech and Keyboard Text Entry for Short Messages in Two Languages on Touchscreen Phones<\/em><\/a>. <\/a>1<\/em><\/a>(4), 23.<\/a><\/p>\n\n\n\n

Russell, S., & Norvig, P. (2002). <\/a>Artificial intelligence: A modern approach<\/em><\/a>.<\/a><\/p>\n\n\n\n

Spielberg, S. (1993). <\/a>Jurassic Park<\/em><\/a>. Universal Pictures. https:\/\/www.imdb.com\/title\/tt0107290\/<\/a><\/p>\n\n\n\n

Starkey, L., Sylvester, A., & Johnstone, D. (2017). Negotiating Digital Divides: Perspectives From the New Zealand Schooling System. Journal of Research on Technology in Education<\/em>, 49<\/em>(1\u20132), 31\u201342. https:\/\/doi.org\/10.1080\/15391523.2017.1292161<\/a><\/p>\n\n\n\n

Starkey, L., & Zhong, J. (2019). The effect of netbook ownership on children\u2019s academic achievement in mathematics, reading, and writing. Journal of Computer Assisted Learning<\/em>, 35<\/em>(1), 34\u201341. https:\/\/doi.org\/10.1111\/jcal.12308<\/a><\/p>\n\n\n\n

Tegmark, M. (2017). <\/a>Life 3.0: Being human in the age of artificial intelligence<\/em><\/a>. Vintage.<\/a><\/p>\n\n\n\n

Togelius, J. (2020, July 17). <\/a>GPT-3 often performs like a clever student who hasn\u2019t done their reading trying to bullshit their way through an exam. Some well-known facts, some half-truths, and some straight lies, strung together in what first looks like a smooth narrative.<\/em><\/a> [Tweet]. Twitter. https:\/\/twitter.com\/togelius\/status\/1284131360857358337<\/a><\/p>\n\n\n\n

Turing, A. M. (1950). I.\u2014COMPUTING MACHINERY AND INTELLIGENCE. <\/a>Mind<\/em><\/a>, <\/a>LIX<\/em><\/a>(236), 433\u2013460. https:\/\/doi.org\/10.1093\/mind\/LIX.236.433<\/a><\/p>\n\n\n\n

Tutt, A. (2017). An FDA for algorithms. <\/a>Admin. L. Rev.<\/em><\/a>, <\/a>69<\/em><\/a>, 83.<\/a><\/p>\n\n\n\n

V\u00f6cking, B., Alt, H., Dietzfelbinger, M., Reischuk, R., Scheideler, C., Vollmer, H., & Wagner, D. (2010). <\/a>Algorithms unplugged<\/em><\/a>. Springer Science & Business Media.<\/a><\/p>\n\n\n\n

V\u00f6cking, B., Alt, H., Dietzfelbinger, M., Reischuk, R., Scheideler, C., Vollmer, H., & Wagner, D. (Eds.). (2011). <\/a>Algorithms Unplugged<\/em><\/a>. Springer Berlin Heidelberg. https:\/\/doi.org\/10.1007\/978-3-642-15328-0<\/a><\/p>\n\n\n\n

 Waluyo, B. (2019). The Effects of ICT on Achievement: Criticizing the Exclusion of ICT from World Bank\u2019s Education Sector Strategy 2020. <\/a>Malaysian Online Journal of Educational Technology<\/em><\/a>, <\/a>7<\/em><\/a>(2), 71\u201387. <\/a>https:\/\/doi.org\/10.17220\/mojet.2019.02.005<\/a><\/a><\/p>\n\n\n\n

Weller, C. (2018). <\/a>Silicon Valley parents are raising their kids tech-free\u2014And it should be a red flag<\/em><\/a>. Business Insider. https:\/\/www.businessinsider.com\/silicon-valley-parents-raising-their-kids-tech-free-red-flag-2018-2<\/a><\/p>\n\n\n\n

Wing, J. M. (2006). Computational thinking. <\/a>Communications of the ACM<\/em><\/a>, <\/a>49<\/em><\/a>(3), 33\u201335.<\/a><\/p>\n\n\n\n

Zubarev, V. (2018). <\/a>Machine Learning for Everyone<\/em><\/a>. Vas3k Blog. http:\/\/vas3k.com\/blog\/machine_learning\/<\/a><\/p>\n\n\n\n

\n\n\n\n

[1]<\/sup><\/a> The italicized text was generated by the author. An NLP system available at https:\/\/textsynth.com\/playground.html<\/a> used the GPT-J 6B model and the words \u201cMachine learning has enabled the spread of fake news\u201d as the input text.<\/p>\n\n\n\n

\n
\n
\n

Current Issues<\/strong><\/h2>\n\n\n\n