AI Influences Society and Society Influences AI
AI’s prevalent and profound benefits are evident from its continued adoption into society and everyday lives. Individuals and organizations increasingly leverage AI technologies and AI-powered systems for tailored recommendations of products and services, content creation, prediction of future purchase habits, and informed decision-making. One of the most common adoptions of AI systems is Netflix’s recommendations on shows and movies to watch based on the audience’s preferences. Besides recommendations, scientists have employed the technologies to create groundbreaking and simultaneously controversial synthetic realities. Wang (2023) defines synthetic realities as digitally produced materials that embody and represent natural environments yet differ significantly from lived experiences. Wide variations of synthetic realities have come up over the years; among the leading ones are deep fake videos such as BuzzFeed’s renditions of former President Obama. Other synthetic materials encompass AI-generated artworks that could pass for human-created, earning the creators loads of money. Such content exemplifies AI’s ability to replicate human experiences. Although the revolutionary AI-powered technologies and resulting excellent content exemplify innovation, personalization, and efficiency, AI presents unprecedented challenges to humans ranging from the explosion of AI-powered technologies to the loss of individuality, the feedback loop of AI-generated and deep-rooted ethical dilemmas, the challenges are profound as they are pervasive.
The media landscape has shifted dramatically because of the prevalence of AI-generated content and persistent user adoption. Numerous content creators like Lil Miquela have amassed a substantial following on social media platforms through AI-generated content or personas. According to Brown et al. (2020), these creators employ technologies like OpenAI’s GPT-3 to generate written content and post on personal blogs or user accounts on social sites. Despite the revolutionary power of AI, it heavily relies on human training to acquire the prerequisite skills to execute its core functions. In turn, it assists humans with a vast knowledge base, creating a loop between AI and humans through their ability to influence each other.
The Explosion of AI-Powered Technologies
The previous decade has been characterized by a dramatic technological shift through the evolution of AI tools and the generation of impressive content in business, education, and healthcare, among several other fields. Castiglioni et al. (2021) state that the advent of machine learning (ML), especially deep learning (DL), has revolutionized content creation. Generative Adversarial Networks (GANs), pioneered in 2014, initiated a novel era that allowed machines to generate images that could be barely distinguished from real ones through generator and discriminator neural networks. The technology’s adversarial training process created deep fakes, which enabled the synthesis of hyper-realistic but completely fabricated video content (Chesney & Citron, 2019). The deep fake algorithms analyze source media to learn patterns, mouth movements, skin textures, and facial expressions, then transpose them onto a target medium, amalgamating realistic individual traits and movements (George & George, 2023). The film industry quickly employed these technologies to de-age their actors, as evidenced by films like “The Irishman” (Wang, 2023). Video makers have also adopted these technologies to create massive, explorable universes tailored to every user’s experience, as seen in the “No Man’s Sky” video game, which leverages procedural generation techniques (Wang, 2023). The proliferation of GANs has resulted in open-source software such as DeepFaceLab and FakeApp that allow anybody to create deepfakes with very little technical know-how.
Today, AI applications and AI-generated content surpass the usage of specialists and extend throughout everyday experiences. For instance, technology-oriented organizations employ AI-powered chatbots to enhance customer experience, while AI-driven frameworks such as GPT-3 aid in writing articles and drafting emails. Individuals and entities leverage AI-powered chatbots in many ways, including scheduling appointments, providing information, booking tickets, and helping people during a crisis, as evidenced by the recent COVID-19 pandemic (Menon & Shilpa, 2023). These programs allow businesses to access wide-ranging information and provide support through text-based conversations powered by natural language processing (NLP) techniques. Commercially, brands leverage AI tools for targeted marketing techniques and creating content customized to their users’ preferences and needs. They exploit AI’s comprehension of theoretical elements of customer engagement and analytical abilities to develop relevant and targeted advertising strategies that promote business growth and more profound customer connection (Menon & Shilpa, 2023). AI-generated content spans the academic world to help students and researchers analyze and create large volumes of data instead of undergoing the traditional laborious manual methods. The recent advancements in NLP technologies have brought about diverse technologies, such as ChatGPT, that leveraged large language models to generate human-like text responses (Babatunde et al., 2024). ChatGPT can generate well-written course essays, answer questions well enough to pass board examinations, summarize research papers, and create helpful computer codes (Van Dis et al., 2023). These examples demonstrate AI’s prevalence across sectors.
In addition to the deepfakes and generated content, AI is perfecting automation efforts in various industries. Cognitive automation is increasingly becoming prevalent, especially in the business environment. According to Grynberg (2019), computers now exceed human beings in several cognition-associated competencies. AI’s cognitive superiority is exemplified in thought-provoking games like chess, checkers, and Go. Even poker is not immune, despite the seemingly built-in attribute of human aspects of the bluff. The advancements in ML allow computers to sort out methods independently to identify patterns in the received data and apply determined rules to the task at hand (Grynberg, 2019). This capability explains why a computer may teach itself to beat a human in a game called Go, which was initially thought to be too complex for AI. However, while AI might be the victor, its internal rules for choosing the most appropriate move may differ from those of a human professional. Also, automation capabilities are increasingly revolutionizing the performance of intense and repetitive tasks through numerical control, computer-aided manufacturing, flexible manufacturing systems, and industrial robots (Javaid et al., 2021). Machines are programmable to execute drills, glass cutting, and 3D printing, while robots aid in wielding assembly and handling materials. Leveraging AI in mundane and intense tasks allows human personnel to focus on more critical operations.
The Erosion of Individuality
The continued progression of AI threatens individuality through a massive influence on human autonomy and decision-making. Rawas (2024) states that AI-driven recommendation systems, social media algorithms, and customized advertising strategies may influence human perceptions, preferences, and behavior, resulting in individual manipulation and persuasion concerns. To this end, balancing protecting human agency and enhancing user experiences becomes paramount in designing and implementing AI systems (Samala & Rawas, 2024). Technical experts and governments must consider the ethical implications of AI-powered manipulated decision-making and enact measures to safeguard citizens from excessive influence. Besides, AI technology developers should embrace ethical directives that spotlight human autonomy while empowering users to make informed decisions and uphold control over their digital experiences.
The increasing advancements in AI technologies intensify the possibility of developing superintelligent or Artificial General Intelligence (AGI) frameworks that augment existential risks (Bucknall & Dori-Hacohen, 2022). Superintelligent systems are AI models that exceed human intelligence across all spheres, prospectively resulting in overwhelming and unexpected repercussions. Creating AGI with precise safety processes and a touch of human values is vital to prevent catastrophic outcomes (Bucknall & Dori-Hacohen, 2022). The fear is that it will overtake human understanding and regulations, leading to unforeseeable actions or choices with unalterable and widespread consequences. Therefore, researchers and policymakers should foster and participate in AGI safety explorations and create global alliances to build regulatory structures prioritizing responsible and safe construction of superintelligent systems.
The Feedback Loop of AI-Generated Knowledge
The feedback loop of AI-generated knowledge manifests as AI feeds the knowledge base, which is now used for new AI recommendations and knowledge creation. Firstly, AI draws from existing datasets to generate content. ML demonstrated how to effectively use data to build efficient algorithms, allowing systems to gain novel knowledge from the datasets. ML can facilitate generative AI to learn new content from vast datasets and create wide-ranging content based on diverse data. For example, in making the Intelligent Tutoring System (ITS), scientists drew from human accounts of solving mathematical problems to train and build accurate frameworks (Cardona et al., 2023). The subsequent system was incorporated into a model to observe students trying to solve mathematical problems on a computer. The scientists who investigated human tutors concluded that feedback on specific steps was essential for optimizing tutoring. They simulated this to the system such that when a student strayed from the expert model, the system afforded them feedback to get them back on track (Cardona et al., 2023). NLP also relies on the comprehension of human language and vast language data to create diverse human-like content. The existing human-based datasets drive the creation of AI-generated data applied in diverse industries, such as education, business, and healthcare.
Secondly, the AI-generated content then enters the knowledge management base. The AI tools use multiple technologies to simulate human intelligence, including deep learning, neural networks, and supervised ML. The highly efficient DL algorithms often assume a supervised approach in which vast loads of labeled data used to train the correlation fortes between nodes in a broad, layered computational network (Armoogum & Li, 2019). As a result, entities employ patterns identified from the data used in training to make precise predictions regarding future unobserved data. This technique differs significantly from classic knowledge management systems like expert systems, which utilize symbolic logic in which humans express and provide the system with rules. Organizations employ the generated data-filled knowledge base to make informed decisions regarding the competitive environment, marketing strategies, operational effectiveness, and customer station initiatives. Even the content on social and professional platforms is becoming increasingly AI-generated (Whittaker et al., 2020). Researchers and content creators rely on AI tools to generate material for posting or publishing, minimizing the laborious manual effort that characterizes traditional research methods. This shift creates an automated knowledge base without human oversight.
Lastly, future AI models use training data from the burgeoning AI-generated knowledge base. Since the automated knowledge base is vastly replacing traditional knowledge bases, thanks to generative AI, scientists and researchers will rely on considerable datasets to train future AI models. AI content is progressively dominating the Internet and institutions, meaning that AI developers will soon have to scrape the Internet for AI-generated content to train new models. The implications of such an undertaking remain unexplored, despite claims from experts that training new AI models with automated datasets will inadvertently bring about errors that accumulate with every ensuing generation of models (Shumailov et al., 2024). Burgeoning evidence backs this assertion, positing that a training dataset of AI-generated material, even in limited amounts, ultimately becomes “poisonous” to the trained framework (Shumailov et al., 2024). For instance, the first generation are language models trained on human datasets. The modes are used to generate AI content, which is then employed to train a novel version of the model whose output trains a third version. Each iteration creates error accumulation that, by the 10th system, a prompt about historical English literature, produces gibberish output about jackrabbits” (Shumailov et al., 2024). Over time, this loop ultimately creates practically meaningless models, skewing the basis of collective understanding.
Historical Parallels
The possibility of AI degradation can be paralleled to the inception of computers or the Internet. Like the current AI democratization efforts, personal computers empowered people and businesses by affording unrivaled access to computer power and revolutionary capabilities. Computers have evolved from the once cumbersome machines limited to specialized environments into becoming a central and universal force fashioning the fabric of modern communities. The benefits surpass technological functionalities to encompass everyday aspects of typical human life, thrusting humanity into a period depicted by unprecedented advancements and innovation. One of the noteworthy benefits of computers is increased productivity. These devices and associated systems have revolutionized businesses and industries through advanced applications and streamlined processes that drive cost-effectiveness and enhance efficiency (Olaoye et al., 2023). Computers also advanced innovation and research through simulation, modeling, and data analysis, accelerating innovation in numerous fields. The inception of personal computers is among the best things to happen to humanity.
The internet boom revolutionized information sharing to augment computers’ functionality and enabled people globally to interact, access substantial amounts of information, and advance research efforts. The Internet drives global interconnectedness through emails, social media, instant messaging, and video conferencing, overcoming geographical barriers and promoting worldwide communication and collaboration (Olaoye et al., 2023). Access to information has created online learning platforms, digital resources, and educational software that have expanded learning opportunities for everyone, regardless of age, background, or other demographic features. The Internet’s role in education and learning was evident during COVID-19 when schools were forced to close to minimize the virus’ spread and rapidly adopted virtual learning from home. Apart from learning and obtaining novel skills, computers and the Internet are fostering research and collaboration among scholars globally through computer-driven communication and sharing findings online.
The transformations brought about by personal computers and the Internet are comparable to current AI revolutions regarding information access. AI takes this a notch further by offering prediction and recommendation abilities based on the generated information. With each invention, humanity achieves new levels of novelty and innovation. Personal computers have changed working machines, the Internet has significantly transformed communication modes among individuals and entities, and AI is continuously transforming industries and professional fields of transportation, finance, healthcare, education, and manufacturing. Every new version of the AI models proves that the technologies may have far-reaching capabilities, some not yet discovered. Humans’ relentless urge to problem-solve and streamline operations drives these innovations and discoveries, some of which may pose considerable ethical implications. For instance, students have been using AI to generate essays and coursework assignments, arguably cheating because it is not the student’s work. To this end, several applications were created to detect AI-generated content and flag the automated work. Still, other software that could modify the AI-generated content to scan as human-generated cropped up, rendering the previous versions invalid. While each version shows significant strides in human ingenuity and AI capabilities, they create an ethical gray area. The increasing overreliance on AI reflects the dependence on computers and the Internet witnessed over the past decades, hampering some vital aspects of human life, such as face-to-face interactions replaced by social media communications. Therefore, these technologies have caused profound societal alterations, augmenting concerns regarding individuality, ethics, knowledge, and originality.
Conclusion
Through generative and predictive capabilities, AI brought groundbreaking solutions to the business environment, healthcare, education, and manufacturing. The generative AI has resulted in mainstream AI-generated content for targeted marketing initiatives on social media platforms, schools, and organization websites. Ai has enabled businesses to make informed decisions, tutors to streamline personalized learning, and content creators to update their pages consistently. However, the prevalence of AI-generated content is threatening individuality because automated responses are replacing people’s unique perspectives, attitudes, and feedback. Besides the loss of individuality, automated content is creating a feedback loop because scientists will soon employ it to train new AI models, creating a poisonous system that will ultimately be rendered meaningless. Therefore, like computers and the Internet have created an overreliance on technologies, AI threatens the core of human authenticity and collective understanding.
References
Armoogum, S., & Li, X. (2019). Big data analytics and deep learning in bioinformatics with hadoop. In Deep learning and parallel computing environment for bioengineering systems (pp. 17-36). Academic Press. https://doi.org/10.1016/B978-0-12-816718-2.00009-9
Babatunde, S. O., Odejide, O. A., Edunjobi, T. E., & Ogundipe, D. O. (2024). The role of AI in marketing personalization: A theoretical exploration of consumer engagement strategies. International Journal of Management & Entrepreneurship Research, 6(3), 936-949. http://dx.doi.org/10.51594/ijmer.v6i3.964
Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J. D., Dhariwal, P., ... & Amodei, D. (2020). Language models are few-shot learners. Advances in neural information processing systems, 33, 1877-1901.
Bucknall, B. S., & Dori-Hacohen, S. (2022, July). Current and near-term AI as a potential existential risk factor. In Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society (pp. 119-129). https://doi.org/10.1145/3514094.3534146
Cardona, M. A., Rodríguez, R. J., & Ishmael, K. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations. Office of Educational Technology.
Castiglioni, I., Rundo, L., Codari, M., Di Leo, G., Salvatore, C., Interlenghi, M., ... & Sardanelli, F. (2021). AI applications to medical images: From machine learning to deep learning. Physica medica, 83, 9-24. https://doi.org/10.1016/j.ejmp.2021.02.006
Chesney, B., & Citron, D. (2019). Deep fakes: A looming challenge for privacy, democracy, and national security. Calif. L. Rev., 107, 1753.
George, A. S., & George, A. H. (2023). Deepfakes: The Evolution of Hyper realistic Media Manipulation. Partners Universal Innovative Research Publication, 1(2), 58-74. http://dx.doi.org/10.5281/zenodo.10148558
Grynberg, M. (2019). AI and the ‘Death of Trademark’. Ky. LJ, 108, 199.
Javaid, M., Haleem, A., Singh, R. P., & Suman, R. (2021). Substantial capabilities of robotics in enhancing industry 4.0 implementation. Cognitive Robotics, 1, 58-75. https://doi.org/10.1016/j.cogr.2021.06.001
Menon, D., & Shilpa, K. (2023). “Chatting with ChatGPT”: Analyzing the factors influencing users’ intention to Use the Open AI’s ChatGPT using the UTAUT model. Heliyon, 9(11). https://doi.org/10.1016%2Fj.heliyon.2023.e20962
Rawas, S. (2024). AI: The future of humanity. Discover Artificial Intelligence, 4(1), 25. https://doi.org/10.1007/s44163-024-00118-3
Samala, A. D., & Rawas, S. (2024). Generative AI as Virtual Healthcare Assistant for Enhancing Patient Care Quality. International Journal of Online & Biomedical Engineering, 20(5). https://doi.org/10.3991/ijoe.v20i05.45937
Shekhar, S. S. (2019). Artificial intelligence in automation. Artificial Intelligence, 3085(06), 14-17.
Shumailov, I., Shumaylov, Z., Zhao, Y., Gal, Y., Papernot, N., & Anderson, R. (2023). The curse of recursion: Training on generated data makes models forget. arXiv preprint arXiv:2305.17493. https://doi.org/10.48550/arXiv.2305.17493
Van Dis, E. A., Bollen, J., Zuidema, W., Van Rooij, R., & Bockting, C. L. (2023). ChatGPT: five priorities for research. Nature, 614(7947), 224-226. https://doi.org/10.1038/d41586-023-00288-7
Wang, Y. (2023). Synthetic realities in the digital age: Navigating the opportunities and challenges of ai-generated content. Authorea Preprints. https://doi.org/10.36227/techrxiv.23968311.v1
Whittaker, L., Kietzmann, T. C., Kietzmann, J., & Dabirian, A. (2020). “All around me are synthetic faces”: the mad world of AI-generated media. IT Professional, 22(5), 90-99. https://doi.org/10.1109/MITP.2020.2985492
