Abstract

Generative Artificial Intelligence (GenAI) is increasingly reshaping a wide range of sectors, including business, healthcare and education, through its ability to generate personalised content and support complex tasks. This paper provides an overview of GenAI’s development from early neural networks to advanced transformer-based models, highlighting its rapid adoption following the release of ChatGPT in 2022. While the benefits of GenAI are substantial – enhancing efficiency, creativity and innovation – its accelerated deployment also raises pressing ethical, social and environmental concerns. These include high energy consumption, electronic waste, privacy breaches, algorithmic bias and the spread of misinformation. Psychological impacts, such as artificial intimacy and overreliance on AI for mental health support, further complicate its use. The paper also considers GenAI’s potential to transform the future of work and support sustainability goals. Ultimately, it calls for a balanced approach to GenAI development – one that fosters innovation while ensuring transparency, fairness and long-term sustainability.

1. Introduction to GenAI

Generative Artificial Intelligence (GenAI) represents a transformative breakthrough within the world of technology. GenAI focuses on generating new creative content – such as text, images, audio, code or video – something that extends far beyond the previous capabilities of existing AIs (Stryker et al., 2025). Unlike traditional AI models, which attempt to distinguish or predict categories within data, generative AI models learn the patterns and relationships within massive datasets and use this knowledge to produce original content in response to prompts or queries (Njoroge, 2025). The emergence of GenAI has enabled people to train these models to learn complex subjects, including human language, programming, art and biochemistry, and apply this understanding to craft innovative outputs that mimic human creativity. 

The most prevalent types of GenAI models include Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs) and modern language models such as Generative Pre-Trained Transformers (GPT) (Lawton, 2025). These models rely on machine learning, utilising neural networks to encode observed data structures in order to generate new, similar content (Njoroge, 2025).

1.1 A BRIEF HISTORY OF GENAI

While today’s strong interest in GenAI, shared by both consumers and businesses alike, was sparked by the rise of ChatGPT in 2022 (Marr, 2023), the technology used by OpenAI to develop the GPT models stems from a long history of computers using neural network algorithms to process datasets and generate outputs from them. In the late 1980s to the 1990s, the AI field advanced with the development of Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks (Bernard, 2023). These networks were able to process sequential data, making them suitable for tasks like speech and language modelling (Marr, 2023). In 2014, this was enhanced with the advent of GANs, pitting two networks – a generator and a discriminator – against each other to generate more high-quality images and video (Stryker et al., 2025). Three years later, the transformer architecture was published, leading to more sophisticated developments in natural language processing, including OpenAI publishing their prototype for the GPT model (ibid.). 

As subsequent models of the GPT were released, many other companies began to invest in their own artificial intelligence language models, such as Google Gemini and Microsoft Copilot, as well as image generators like DALL-E. The release and global domination of ChatGPT has demonstrated GenAI’s ability to generate consistent, context-aware and human-like language at scale, propelling AI into mainstream awareness and use.

1.2 The Technical Aspect of GenAI

AI works by enabling computational systems to mimic tasks associated with human intelligence through several key technical components and processes (Boucher, 2020). At its core, AI operates by ingesting data, processing it through models, primarily artificial neural networks (ANNs), and learning from the data to make predictions or decisions with minimal human intervention. AI systems function through data inputs and data processing. AI systems will collect large and varied datasets (e.g. text audio and video) and prepared protocols (Drexel et al., 2019). The dataset used for the AI is crucial as it defines the quality and scope of the information the AI can learn from.

1.2.1 ARTIFICIAL NEURAL NETWORKS

AI learning and processing is modelled after systems of neural networks. ANNs are modelled loosely on the way the human brain records and perceives information (Scott, 2025). They consist of nodes known as artificial neurons, arranged in layers: an input layer, one or more hidden layers, and an output layer (see Figure 1). These networks work in tandem to transmit data, with each node applying a function to transform the data (Marwala, 2024). The more complex the AI model is, the more hidden layers it consists of, allowing the neural network to progressively extract as many features as possible from the raw dataset (Craig et al., 2024).

Figure 1: A System of Artificial Neural Networks (Rahman, 2024, p.3).

1.2.2 AI LEARNING ALGORITHMS

AI systems learn by adjusting the neural network’s internal weights to reduce the error between their output predictions and actual outcomes. Back-propagation (see Figure 2) is a method where the system calculates the error at the output layer and propagates this error backward through the network’s layers to update the neurons’ weights, improving accuracy with each iteration (Scott, 2025). AI models may also use gradient descent, an optimisation technique that efficiently searches the space of possible network configurations to find the set of weights minimising the prediction error, guiding the learning process (Craig et al., 2024).

Figure 2: A Diagram of a Back-Propagation Model (Leonel, 2018, p.1).

1.2.3 MACHINE LEARNING & DEEP LEARNING

Machine learning automates the learning process by enabling the network to adjust itself from data without explicit programming of rules (Marwala, 2024). Deep learning is a subset that uses deep neural networks for enhanced pattern recognition capable of complex tasks like image recognition, natural language processing and speech recognition (Scott, 2025).

1.2.4 GENERATIVE MODELS & LARGE LANGUAGE MODELS

Advanced AI, such as GenAI (e.g. ChatGPT), uses large language models trained on massive datasets with deep learning to generate new content by predicting the next element in a sequence, like words in a sentence (Boucher, 2020). This capability is an extension of pattern recognition into creative generation.

1.3 Advantages of GenAI

As GenAI moves closer to the forefront of the public consciousness, industries and individuals aim to adopt GenAI into their everyday awareness. This section analyses case studies to identify and understand the key benefits that GenAI is able to provide in its current state. 

1.3.1 QUALITY OF WORK

Tools like ChatGPT can assist with problem formulation, research design, data collection and analysis, as well as reviewing and critiquing writing and composition (Susarla et al., 2023). In addition, there is a perception in the improvement in internal workplace communications in terms of professionalism, clarity and efficiency among employees (Cardon et al., 2023). These factors understandably improve the quality of work produced in the corporate field, increasing revenue. For these reasons, those in managerial or executive positions tend to be more enthusiastic about the use of GenAI in the workplace than those in non-managerial positions (Cardon et al., 2023). 

1.3.2 CONTENT GENERATION

Coman and Cardon’s (2024) study suggests that people value GenAI’s content generation abilities more than its editing assistance capabilities, particularly in creative writing tasks.

1.3.3 AUTOMATION & PRODUCTIVITY

AI systems hold promise to deliver efficiency and increase human productivity. On a study conducted on GenAI in the software development field, findings conclude that the impact of GenAI was significant, with the group that utilised Github Copilot completing the task 55.8% faster than the control group, with most skill levels being able to benefit from the tool (Peng et al., 2023). 

Automation through GenAI has seen use particularly through industry. Easily automatable industries include accommodation and food services, manufacturing, transportation, warehousing and retail trade (Manyika et al., 2017). Indeed, AI-powered systems are seeing use in warehouses, enabling automated inventory tracking and optimisation, AI-powered robotics taking on labour-intensive tasks like order-picking, in-house navigation, retrieving products from warehouse shelves and delivery (Sodiya et al., 2024)

1.3.4 ACCESSIBILITY

Glazko et al. (2023) recognise the potential for the use of GenAI as an accessibility tool, such as using its summarisation and information extraction capabilities to aid those with information processing difficulties, or its visual imagery creation abilities to aid in those with visualisation difficulties. The appeal of this approach stems from its ability to provide immediate on-demand support for accessibility needs on an individual level. The use of GenAI has shown success in a collection of use cases spanning individuals possessing attention deficit hyperactivity disorder, generalised anxiety disorder, autism and bipolar disorder (Clark, 2023).

1.3.5 PERSONALISATION

GenAI’s capability of generating novel and unique content based on existing information has been subject to use in marketing and advertising. GenAI is able to efficiently analyse and identify patterns in consumer behaviour, generate highly personalised data and provide real-time engagement with audiences (Gujar and Panyam, 2024). 

1.4 Challenges in GenAI

As GenAI technologies continue to revolutionise sectors such as business, healthcare and education, they also raise complex challenges that transcend technological performance. Key concerns include environmental sustainability, psychological effects, cultural-linguistic equity, privacy, algorithmic fairness and regulatory gaps. This section outlines some of the most pressing societal and ecological implications associated with GenAI deployment.

1.4.1 ENVIRONMENTAL CHALLENGES

The rapid expansion of GenAI technologies has sparked growing concern over their environmental sustainability. One of the most pressing issues is the massive energy and water consumption associated with training and operating large language models (LLMs). For example, training a single large-scale GenAI model can consume hundreds of megawatt-hours of electricity and millions of litres of freshwater (Ming et al., 2024). These energy demands not only increase greenhouse gas emissions but also contribute to the depletion of limited freshwater reserves, a problem intensified by geographic inequalities in data centre locations (Raji et al., 2024).

Another key environmental concern is electronic waste (e-waste), resulting from the frequent upgrading and discarding of GPUs, CPUs and supporting infrastructure used in GenAI development (Ming et al., 2024). This contributes to the broader crisis of global digital pollution and toxic material accumulation, especially in low- and middle-income countries that bear the brunt of e-waste processing (Raji et al., 2024). Additionally, recent life cycle assessment (LCA) methodologies have revealed that even seemingly “intelligent” applications like GenAI-powered image generators can have a disproportionately high carbon footprint throughout their entire operational lifespan (Osseiran et al., 2024). As the technology continues to evolve, integrating environmental assessments into AI development lifecycles will be crucial to reducing these ecological impacts.

1.4.2 PSYCHOLOGICAL IMPACT

The rapid shifting in human-AI relationship paradigm might raise concerns about mental health, emotional wellbeing and ethical boundaries. As AI systems increasingly mimic human interaction, users form deep emotional connections with chatbots and virtual agents, a phenomenon now studied under the term “artificial intimacy” (Schoen & Hancock, 2024). These relationships can range from friendships to romantic partnerships and even extend to users expressing desires to raise children with AI companions (Daily Star, 2023). Notably, GenAI is also being used as a surrogate therapist. While AI-based mental health tools offer accessibility and privacy, they lack the accountability, empathy and nuance of licensed professionals (Bedi et al., 2019). Overreliance on AI for psychological support may lead to self-isolation or the deferral of professional care, especially among vulnerable users.

Moreover, widespread surveys have shown that a significant portion of the public is open to emotional or even physical relationships with humanoid robots, revealing shifting cultural norms around intimacy and ethics in human-machine interaction (Daily Star, 2023). These developments underscore the need for interdisciplinary research on the psychological effects of AI and the implementation of guardrails to ensure ethical use.

1.4.3 LINGUISTIC & CULTURAL CHALLENGES

Language and cultural bias present another significant challenge for GenAI. Many of today’s LLMs are trained predominantly on English-language datasets, which not only disadvantages non-English speakers but also risks epistemic injustice by excluding local knowledge systems, dialects and Indigenous languages (Cerrato & Ciuccarelli, 2023). This anglocentric bias can marginalise diverse voices and reinforce global power imbalances in knowledge production and dissemination.

Additionally, AI-enabled systems deployed in recruitment, content moderation and public services often exhibit biases against linguistic minorities, reinforcing structural discrimination (Thompson et al., 2023). Such systems may fail to interpret meaning in culturally-specific expressions or undervalue qualifications from non-English academic systems, leading to exclusion and inequality. Addressing these issues requires not only broader and more inclusive training corpora but also a critical re-evaluation of the linguistic assumptions embedded in AI architectures.

1.4.4 PRIVACY

GenAI models are trained on datasets that are scraped from the internet; these datasets often contain personal or sensitive information. Despite efforts to anonymise datasets, Carlini et al. (2021) show that these models can still memorise and reproduce specific data points under particular prompts, thereby inadvertently compromising user privacy, unlike traditional big data analytics, which primarily query static datasets. GenAI’s dynamic ability to regenerate content makes it difficult to trace the origin of privacy violations. This poses a unique challenge in regulating and ensuring the protection of sensitive information (Al-kfairy et al., 2024).

1.4.5 BIAS & FAIRNESS

Bias remains a big concern within GenAI systems, largely due to the biases inherent in training data. As noted by Bender et al. (2021), LLMs can perpetuate harmful stereotypes related to race, gender and other factors, often reflecting societal inequalities embedded in the data. These biases are particularly problematic in areas like hiring, criminal justice and education, where GenAI systems may reinforce discrimination without accountability. Wei et al. (2025) suggest that addressing these biases requires comprehensive strategies, such as dataset curation, bias auditing and the development of fairness metrics to reduce algorithmic harm and ensure equitable outcomes.

1.4.6 MISINFORMATION & POLICY

The generative capabilities of AI models also raise concerns about misinformation. These systems are known to produce “hallucinations” , realistic-sounding but incorrect information (Bommasani et al., 2022). The growing reliance on AI-generated content necessitates robust regulatory frameworks to address the risks of misinformation; however, Christodorescu et al. (2024) state that current policies, such as the EU AI Act and the U.S. Executive Order on AI, have yet to catch up with the rapid advancements in GenAI. This regulatory gap creates challenges in ensuring transparency, accountability and the prevention of misuse.

To sum up, GenAI presents major societal and environmental challenges alongside its benefits. Key concerns include high energy and water use in model training, electronic waste and a significant carbon footprint (Ming et al., 2024; Osseiran et al., 2024). Psychologically, GenAI fosters artificial intimacy and raises mental health risks due to overreliance on AI companions (Schoen & Hancock, 2024; Bedi et al., 2019). Linguistic and cultural biases exclude non-English speakers and marginalise local knowledge (Cerrato & Ciuccarelli, 2023). GenAI also risks privacy breaches through data memorisation (Carlini et al., 2021) and perpetuates bias in decision-making systems (Bender et al., 2021; Wei et al., 2025). Additionally, its capacity to generate misinformation highlights the urgency for effective regulation (Bommasani et al., 2022; Christodorescu et al., 2024).

2. GenAI in Different Sectors

GenAI is rapidly transforming different types of industries. This section explores the diverse applications and impacts of GenAI across key sectors, including education, business, marketing and healthcare, highlighting its transformative potential and the sector-specific considerations for its responsible deployment.

2.1 GenAI in Education

GenAI is increasingly transforming education by enabling new approaches to teaching and learning. GenAI refers to AI systems capable of creating new content, such as text or images in response to prompts, with examples including large language models like ChatGPT (Chan & Hu, 2023). In educational settings, these technologies offer opportunities for more personalised learning experiences and efficient content delivery. Early integrations of GenAI in classrooms have shown its potential to provide individualised support and instant feedback to students, thereby enhancing engagement (Garcia-López & Trujillo-Liñán, 2025). 

GenAI’s main benefit in education is its ability to deliver personalised learning and feedback at scale. It can adapt to individual students’ needs, providing tailored explanations, examples and hints, which aids in understanding difficult concepts outside of regular classroom hours (Liu et al., 2024). Studies show that GenAI tools like chatbots can enhance engagement and motivation, improving skills like argumentation in debates (Guo et al., 2023). Additionally, university students report that generative AI aids their learning by offering writing, brainstorming and research support, benefitting non-native speakers and those needing extra practice (Chan & Hu, 2023). 

Educators can also stand to benefit from GenAI through automated content creation and grading assistance. GenAI systems can generate quiz questions, summaries of reading materials or feedback on assignments, thereby reducing the workload on instructors for routine tasks. Early experiments indicate that such tools can free up teachers’ time to focus on higher-level instructional design and one-on-one mentoring (Garcia-López & Trujillo-Liñán, 2025). In addition, GenAI can analyse large amounts of student data (e.g. past performance or common errors) and provide insights to educators, helping them tailor their teaching strategies. Overall, the promise of GenAI in education lies in enhancing personalisation and scalability: it can deliver adaptive learning experiences where content and support adjust in real time to each student, something traditional classrooms often struggle to achieve.

Universities are also utilising genAI in the classroom. For example, Stanford University has been a leader in testing GenAI within higher education (Garcia-López & Trujillo-Liñán, 2025). In 2023, the university launched initiatives to integrate AI models like ChatGPT into several courses as a means to enhance learning (ibid.). Rather than replacing instructors or human tutors, the GenAI tools at Stanford were used as supportive aids; for instance, helping students brainstorm ideas, refine their writing or engage in critical thinking exercises. Early reports from these trials showed that students benefitted from more interactive and personalised learning activities, as the AI could respond to their queries with detailed explanations or examples on demand. However, the Stanford case also underscored certain challenges. University educators observed the need to teach digital literacy skills so that students could effectively use AI outputs, such as learning to verify the accuracy of AI-generated content and not accept it uncritically (Garcia-López & Trujillo-Liñán, 2025). There were also efforts to develop clear ethical guidelines for GenAI use in coursework, emphasising academic integrity and responsible use of AI assistance (ibid.). This case study illustrates that while GenAI can improve engagement and provide tailored support, institutions must establish strong frameworks (e.g. honour codes or usage policies) to guide its ethical and pedagogical implementation (ibid.).

2.2 GenAI in Business and Marketing

GenAI is rapidly transforming business operations and marketing strategies. According to Woolley (2024), businesses are leveraging GenAI for content generation, personalisation and automation, thereby improving operational efficiency and strategic insight. Bi (2023) highlights a surge in GenAI adoption in marketing, with firms automating repetitive tasks and using these systems to craft personalised campaigns at scale. Further, Singh et al. (2024) suggest that GenAI is redefining customer interaction by enabling automated yet context-aware marketing agents that manage consumer touchpoints throughout the customer journey.

2.2.1 CONTENT CREATION & PERSONALISATION

A key application of GenAI in marketing is automated content creation. Firms employ tools like large LLMs to generate email campaigns, social media posts and promotional material rapidly and with minimal human input (Bi, 2023; Woolley, 2024). This automation not only reduces operational costs but also enhances message consistency. LLMs can tailor content to specific customer segments, producing personaliSed communications that resonate more effectively with audiences. Aghaei et al. (2025) support this by demonstrating significant improvements in customer engagement when AI-generated marketing is precisely targeted.

2.2.2 INTELLIGENT CUSTOMER ENGAGEMENT

GenAI facilitates deeper customer insights through data-driven personaliSation. Bi (2023) documents how GenAI-powered agents analyse user behaviour to segment audiences, predict purchasing intentions and generate hyper-personaliSed outreach, thereby optimising campaign effectiveness and conversion rates. Singh et al. (2024) describe “AI marketing agents” as systems capable of autonomously interacting with customers, responding to inquiries and adapting promotions based on real-time user data.

2.2.3 STRATEGIC DECISION SUPPORT

Beyond external marketing efforts, GenAI aids internal decision-making. Automated systems assist in budget allocation, scenario modelling and performance monitoring (Woolley, 2024). Aghaei et al. (2025) emphasise that real-time predictive analytics empower marketing teams to pivot strategies based on emerging consumer trends, thereby enhancing agility and competitive responsiveness.

2.2.4 CHALLENGES: BIAS, ETHICS & TRUST

Despite its advantages, GenAI poses significant challenges, particularly regarding algorithmic bias, ethical use and transparency. Researchers warn that if training data reflects systemic biases, AI-generated outputs may reinforce stereotypes or marginaliSe demographic groups (Bi, 2023). According to Singh et al. (2024), biased messaging not only damages brand equity but also erodes customer trust and invites regulatory scrutiny. Woolley (2024) notes that ethical compliance and transparency are essential for maintaining public confidence – without clear disclosure of AI-generated content, consumer trust may decline.

Overall, GenAI delivers transformative opportunities for content creation, personalisation and decision support in business and marketing. However, institutions should address bias, maintain ethical standards and communicate transparently to preserve stakeholder trust. As businesses continue to pilot GenAI tools, monitoring their impact on brand reputation and regulatory compliance is essential. The balance between innovation and responsibility will determine whether GenAI strengthens or undermines future business outcomes.

2.3 GenAI in Healthcare 

The global healthcare industry has been at the forefront of embracing Gen AI. The industry has been an early adopter of this technology and has helped build trust in GenAI (Jain, 2025). Healthcare providers across the world are widely using GenAI in real-world situations (ibid.). It has supported increased personalisation of healthcare and improved efficiency.

2.3.1 CLINICAL APPLICATIONS

There are many existing GenAI applications in use. For example, GenAI can enhance medical imaging by producing high-resolution images from low-quality scans, enabling easier analysis and diagnosis for healthcare professionals (Jain, 2025). In addition, GenAI algorithms are already streamlining drug development by designing new molecules with targeted properties, significantly reducing both time and costs associated with traditional pharmaceutical research (The Lancet, 2023). 

By processing large datasets, GenAI can find patterns within data and predict high risk patients, even before they or their doctors can identify any symptoms of illness or disease (Rajkomar et al., 2019). Gen AI also has the ability to create synthetic data which solves issues with privacy of patient data (The Lancet, 2023). Synthetic data is created from anonymised existing patient data. Previously real patient data would have to be used for education and training purposes; now synthetic data can be used (ibid.).

2.3.2 PERSONALISATION

There is a very significant opportunity to improve healthcare worldwide through GenAI personalised healthcare. GenAI is able to analyse large datasets with multidisciplinary patient information and wider public health data in order to provide patients with comprehensive and personalised care plans (Rajkomar et al., 2019). Historic patient records, test results, genetic information and potential treatments can all be consolidated to create data-informed, detailed and personalised plans. GenAI platforms are able to do this much more efficiently, compared to a human creating a care plan where they do not have the time or ability to analyse these large datasets (Baig et al., 2024).  

On the other hand, the risks of such an approach should be considered, especially with the lower amount of research undertaken in minority groups. Personalised treatments may not be accurate if data is skewed or there is not enough data from minority groups. In order to develop personalised medicine, GenAI needs to process existing patient data. It is crucial that this data is both “diverse and representative” (Shaban-Nejad et al., 2023). Shaban-Nejad et al. (2023) state that if the training data is skewed, it may not generalise well to under-represented groups.

2.3.3 INCREASED EFFICIENCY

GenAI has huge potential to transform healthcare through increasing efficiency by processing and interpreting the substantial amount of pre-existing data (Zhang & Kamel Boulos, 2023). However, there is the opportunity for improved patient outcomes across many areas of healthcare through more efficient and data-focused diagnosis.   

GenAI could have a significant positive impact on reducing the administrative time of doctors. McNally and Huber (2021) estimated that doctors spend 44% of their time on administration (McNally & Huber, 2021). Clinical notes, treatment summaries and referral letters could all become more efficient while also offering collaboration opportunities with multidisciplinary teams. AI tools, such as Nuance Dragon Medical and Suki AI, can reduce documentation time by up to 50% and achieve transcription accuracy of 95% (Batista, 2025). Through automatic speech recognition, Suki AI can transcribe a conversation between a doctor and patient into note form, improving the quality of communication between both parties and therefore improving patient engagement and empowerment. The potential to reduce the administrative burden on highly trained and specialist doctors would allow more time to diagnose and treat patients, ultimately reducing waiting times and improving patient outcomes.

2.3.4 PUBLIC HEALTH APPLICATIONS

There is work ongoing for the application of GenAI for the planning, preparedness and response to a pandemic or epidemic. During the COVID-19 pandemic, modelling, analysing and aggregating test data and reporting was mostly manual, making the processes slower than the spread of disease (Hassan, 2024). Speeding up these processes through GenAI and the ability to make wide-reaching data-driven decisions could have a significant impact on reducing and stopping future pandemics. These methods could be applied to other public health situations, such as modelling and reporting on risk areas for environmental health concerns (Hassan, 2024).

3. The Future of GenAI

GenAI is already making huge changes to how we work and changing the shape of the labour market by automating certain sectors of the employment market. While there are enormous opportunities for efficiency, it is essential that the right checks and balances are in place for a positive future working together with GenAI.

3.1 The future of how we work

GenAI is certain to increase the productivity of many jobs as it can be used for problem solving, decision-making and predicting changes (Cazzaniga et al., 2024). This means there is a potential positive impact on many people working today as it can assist with and improve accuracy of tasks and projects. However, there are a vast number of applications to existing jobs, and we are not yet able to predict the impact it will have overall on the economy or society (ibid.).

It was previously thought that lower-skilled or task-orientated jobs only would be replaced, and higher skilled jobs with advanced problem-solving and reasoning skills would not be replaced by AI (Cazzaniga et al., 2024). However, due to the rapid advances in GenAI and the current and future development that is taking place, these highly-skilled jobs are also at risk of being replaced as GenAI becomes more efficient and can analyse vast amounts of data and problem solve (ibid.). There is great potential for GenAI to automate previously manual administrative tasks (Joshi, 2024). Although this could save on labour costs for organisations, this puts these manual and administrative jobs at risk. There is, however, potential for GenAI to create new jobs in areas such as AI development, data analysis and creative industries (ibid.). There might be an increased need for AI development jobs due to its rapid growth (Cazzaniga et al., 2024).

3.2 Human-AI collaboration

As use of GenAI becomes increasingly mainstream, it is important to consider the future of human-AI collaboration. A Human Centred Artificial Intelligence (HCAI) framework is recommended by Shneiderman (2020) which argues for high levels of computer automation alongside high levels of human control, thus increasing human output without relinquishing control or responsibility.

3.3 Environmental Considerations

It is essential that a responsible AI future is created, and this is one of the greatest challenges. Gen AI ultimately has a negative impact on the environment, however it can also be used effectively to help optimise energy use and efficiency of electrical devices and energy production (Berthelot, 2024). Indeed, GenAI has an enormous environmental impact, specifically, increasing energy consumption through processing and also contributing to metal scarcity due to the hardware required (ibid.). However, it also has the ability to help us progress with renewable energy (Gade, 2024). 

GenAI models can make a big difference in improving renewable energy planning. GenAI has the ability to analyse renewable energy networks, like wind and solar power, to work as efficiently as possible, and the ability to analyse current or future networks to see how much energy they can provide, supporting planning for other energy sources (Gade, 2024). By optimising global renewable energy infrastructure, we would hope to reduce the current fossil fuels that are being released (ibid.).

4. Conclusion

The rapid rise of GenAI over the last three years has presented many opportunities and challenges. The launch of ChatGPT in 2022 sparked global interest in GenAI, although the foundations for these models were laid as early as the 1980s. The incredibly fast global adoption of a new technology has presented many challenges. For example, the intense energy requirements of GenAI have set back environmental targets. There has also been a rise in artificial intimacy, with a concern that this can lead to self-isolation and a reliance on GenAI for mental health support without the necessary ethical and regulatory safety measures in place. There are real concerns around privacy of the large datasets that GenAI uses and this is an area that still needs much work and potentially regulation. Bias is prevalent in many GenAI models due to skews in the training data, which will take thorough strategies to eliminate in the future.  

On the other hand, there are clear opportunities. In education, GenAI offers personalised learning and feedback, while reducing pressure on educators. This personalised approach also extends to marketing and customer engagement, while also increasing efficiency. The healthcare industry has been an early adopter of GenAI, with impressive examples of increased accuracy and efficiency.

There is huge potential for the future of GenAI, but it is essential that we address both the key opportunities and challenges thoroughly to ensure that this new technology is used both responsibly and ambitiously.

Aghaei, R., Kiaei, A.A., Boush, M., Vahidi, J., Zavvar, M., Barzegar, Z. & Rofoosheh, M. (2025) Harnessing the Potential of Large Language Models in Modern Marketing Management: Applications, Future Directions, and Strategic Recommendations. arXiv, 2501.10685.

Ali, H. (2024) AI for pandemic preparedness and infectious disease surveillance: predicting outbreaks, modeling transmission, and optimizing public health interventions. Int J Res Publ Rev, 5(8), pp.4605-19.  

Al-kfairy, M., Mustafa, D., Kshetri, N., Insiew, M.M. & Alfandi, O. (2024) Ethical Challenges and Solutions of Generative AI: An Interdisciplinary Perspective. Available at: https://www.mdpi.com/2227-9709/11/3/58?ref=compliancehub.wiki

Baig, M.M., Hobson, C., GholamHosseini, H., Ullah, E. & Afifi, S. (2024) Generative AI in Improving Personalized Patient Care Plans: Opportunities and Barriers Towards Its Wider Adoption. Applied Sciences, 14(23), p.10899. doi:https://doi.org/10.3390/app142310899

Batista, A.S. (n.d.) Ai in healthtech: Building HIPAA-compliant solutions for next-generation medical documentation.  Journal of Next-Generation Research 5.0. Available at: https://jngr5.com/index.php/journal-of-next-generation-resea/article/view/124

Bedi, G., Carrillo, F., Cecchi, G.A., Slezak, D.F., Sigman, M., Mota, N.B., Ribeiro, S., Javitt, D.C., Copelli, M. & Corcoran, C.M. (2019) Your robot therapist will see you now: Ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy. Journal of Medical Internet Research, 21(5), p.e13216. Available at: https://www.jmir.org/2019/5/e13216/ 

Bender, E.M., Gebru, T., McMillan-Major, A. & Mitchell, M. (2021) On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? Available at: https://dl.acm.org/doi/abs/10.1145/3442188.3445922 

Berthelot, A., Caron, E., Jay, M. & Lefevre, L. (2024) Estimating the environmental impact of Generative-AI services using an LCA-based methodology. Procedia CIRP, 122, pp.707–712. doi:https://doi.org/10.1016/j.procir.2024.01.098

Bi, Q. (2023) Analysis of the application of generative AI in business management. Advances in Economics and Management Research, 6(1), p.36.

Boucher, P. (2020) Artificial Intelligence: How Does It Work, Why Does It Matter, and What Can We Do about It? Available from: https://www.europarl.europa.eu/RegData/etudes/STUD/2020/641547/EPRS_STU(2020)641547_EN.pdf

Bommasani, R. et al. (2022) On the Opportunities and Risks of Foundation Models. Stanford CRFM Report. arXiv.

Cardon, P.W., Getchell, K., Carradini, S., Fleischmann, C. & Stapp, J. (2023) Generative AI in the workplace: Employee perspectives of ChatGPT benefits and organizational policies. SocArXiv.

Cazzaniga, M. et al. (2024) Gen-ai: Artificial Intelligence and the future of work. IMF. Available at: https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues/2024/01/14/Gen-AI-Artificial-Intelligence-and-the-Future-of-Work-542379

Cerrato, A. & Ciuccarelli, P. (2023) Diversity and Language Technology: How Techno-Linguistic Bias Can Cause Epistemic Injustice. arXiv. Available at: https://arxiv.org/abs/2307.13714 

Chan, C.K.Y. & Hu, W. (2023) Students’ voices on generative AI: perceptions, benefits, and challenges in higher education. International Journal of Educational Technology in Higher Education, 20(43). doi:https://doi.org/10.1186/s41239-023-00411-8

Chatterjee, Samir, et al. (2024) Towards New Frontiers of Healthcare Systems Research Using Artificial Intelligence and Generative AI. Health Systems, 13(4), pp.263–273. doi:https://doi.org/10.1080/20476965.2024.2402128

Clark, M. (2023) AI, generative production, & accessibility. Teaching and Learning at Cleveland State University.

Coman, A.W. & Cardon, P. (2024) Perceptions of professionalism and authenticity in AI-assisted writing. Business and Professional Communication Quarterly, p.23294906241233224.

Drexl, J., Hilty, R., Beneke, F., Desaunettes-Barbero, L., Finck, M. et al. (2019) Technical Aspects of Artificial Intelligence: An Understanding from an Intellectual Property Perspective. SSRN Electronic Journal. doi:https://doi.org/10.2139/ssrn.3465577

Daily Star (2023) Huge number of people admit they’d be cool with their partner hooking up with a robot. Available at: https://www.dailystar.co.uk/news/weird-news/huge-number-people-admit-theyd-34813456

Gade, P.K. (2019) MLOps Pipelines for GenAI in Renewable Energy: Enhancing Environmental Efficiency and Innovation. Asia Pacific Journal of Energy and Environment, 6(2), pp.113–122. doi:https://doi.org/10.18034/apjee.v6i2.776

Garcia-López, I.M. & Trujillo-Liñán, L. (2025) Ethical and regulatory challenges of Generative AI in education: a systematic review. Frontiers in Education, 10, 1565938. doi:https://doi.org/10.3389/feduc.2025.1565938.

Glazko, K.S., Yamagami, M., Desai, A., Mack, K.A., Potluri, V., Xu, X. & Mankoff, J. (2023) An autoethnographic case study of generative artificial intelligence’s utility for accessibility. Proceedings of the 25th International ACM SIGACCESS Conference on Computers and Accessibility, pp. 1-8.

Gujar, P. & Panyam, S. (2024) Generative AI in digital advertising campaigns. International Journal of Computer Trends and Technology, 72(5), pp.51-55.

Guo, K., Zhong, Y., Li, D. & Chu, S.K.W. (2023) Effects of chatbot-assisted in-class debates on students’ argumentation skills and task motivation. Computers & Education, 203, 104862. doi:https://doi.org/10.1016/j.compedu.2023.104862

Jain, A., Rao, A.S. & Agarwal, S. (2025) Exploring the Role of Generative AI in Modern Healthcare: Innovations, Applications, and Future Prospects. PRATIBODH (NCICT).

Joshi, S. (2025) Introduction to Generative AI: Its Impact on Jobs, Education, Work and Policy Making. Introduction to Generative AI: Its Impact on Jobs, Education, Work and Policy Making, (2). doi:https://doi.org/10.20944/preprints202503.2126.v2

Lawton, G. (2025) What Is GenAI? Generative AI Explained. TechTarget. Available at: https://www.techtarget.com/searchenterpriseai/definition/generative-AI 

Leonel, J. (2018) Backpropagation. Medium. Available at: https://medium.com/@jorgesleonel/backpropagation-cc81e9c772fd

Lev, C., Laskowski, N. & Tucci, L. (2024) What Is Artificial Intelligence (AI)? TechTarget. Available at: https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence 

Liu, S., Guo, X., Hu, X. & Zhao, X. (2024) Advancing generative intelligent tutoring systems with GPT-4: design, evaluation, and a modular framework for future learning platforms. Electronics, 13(24), p.4876. doi:https://doi.org/10.3390/electronics13244876

Manyika, J., Chui, M., Miremadi, M., Bughin, J., George, K., Willmott, P. & Dewhurst, M. (2017) A future that works: AI, automation, employment, and productivity. McKinsey Global Institute Research (60), pp.1-135.

Marr, B. (2023) A Simple Guide to the History of Generative AI. Bernard Marr and Co. Available at: https://bernardmarr.com/a-simple-guide-to-the-history-of-generative-ai/ 

Marwala, T. (2024) AI Is Not a High-Precision Technology, and This Has Profound Implications for the World of Work. United Nations University. Available at: https://unu.edu/article/ai-not-high-precision-technology-and-has-profound-implications-world-work 

McNally, S. & Huber, J. (2021) Developing a ‘doctors’ assistant’ role to ease pressure on doctors and improve patient flow in acute NHS Hospitals. BMJ Leader. Available at: https://bmjleader.bmj.com/content/5/1/62 

Ming, R., Lin, Z., Paleyes, A., Hu, R., Li, L., Zhu, X., Ge, S., Deng, H. & Van der Schaar, M. (2024) Estimating the environmental impact of Generative-AI services using an LCA-based methodology. Sustainable Computing: Informatics and Systems. Available at: https://www.sciencedirect.com/science/article/pii/S2212827124001173 

Njoroge, J. (2025) Evolution of Generative AI: Shaping Tech Solutions. Entrans. Available at: https://www.entrans.ai/blog/evolution-of-generative-ai 

Osseiran, A., Trancoso, I. & Li, X. (2024) A social-environmental impact perspective of generative artificial intelligence. Patterns, 5(6), 100949. Available at: https://www.sciencedirect.com/science/article/pii/S2666498424001340 

Peng, S., Kalliamvakou, E., Cihon, P. & Demirer, M. (2023) The impact of ai on developer productivity: Evidence from github copilot. arXiv. 2302.06590.

Rahman, J. (2024) Simply Understanding Aritifical Neural Networks (ANN): Deep Learning. Medium. Available at: https://medium.com/@jeslurrahman/simply-understanding-artificial-neural-networks-ann-deep-learning-10b19e2eadca

Raji, I.D., Fielkow, C., Grobe, C. & Ahmad, A. (2024) Environmental footprint of GenAI – Changing technological future or planet climate? Journal of Responsible Technology, 18, 100107. Available at: https://www.sciencedirect.com/science/article/pii/S2444569X25000411 

Rajkomar, A., Dean, J. * Kohane, I. (2019) Machine Learning in Medicine. New England Journal of Medicine, 380(14), pp.1347–1358. doi:https://doi.org/10.1056/nejmra1814259

Schoen, H. & Hancock, J.T. (2024) Artificial Intimacy: Understanding emotional bonds with conversational agents. Journal of Human–Computer Interaction, 40(2), pp.105–121.

Scott, G. (2025) Artificial Intelligence (AI): What It Is, How It Works, Types, and Uses. Investopedia. Available at: https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp 

Shaban-Nejad, A. Michalowski, M. & Bianco, S. (2023) Creative and generative artificial intelligence for personalized medicine and healthcare: Hype, reality, or hyperreality? Experimental biology and medicine, 248(24), pp.2497–2499. doi:https://doi.org/10.1177/15353702241226801

Shneiderman, B. (2020) Human-centered artificial intelligence: Reliable, safe trustworthy. International Journal of Human–Computer Interaction, 36(6), pp.495–504. doi:https://doi.org/10.1080/10447318.2020.1741118

Singh, N., Chaudhary, V., Singh, N., Soni, N. & Kapoor, A. (2024) Transforming business with generative ai: Research, innovation, market deployment and future shifts in business models. arXiv, 2411.14437.

Sodiya, E.O., Umoga, U.J., Amoo, O.O. & Atadoga, A. (2024) AI-driven warehouse automation: A comprehensive review of systems. GSC Advanced Research and Reviews, 18(2), pp.272-282.

Stryker, C. & Scapicchio, M. (2024) What Is Generative AI? IBM. Available at: https://www.ibm.com/think/topics/generative-ai 

The Lancet Regional Health (2023). Embracing generative AI in health care. The Lancet Regional Health, 30, pp.100677–100677. doi:https://doi.org/10.1016/j.lanepe.2023.100677.  

Thompson, B., Madgavkar, A. & Kennedy, B. (2023) Ethics and discrimination in artificial intelligence-enabled recruitment practices. Humanities and Social Sciences Communications, 10(103). Available at: https://www.nature.com/articles/s41599-023-02079-x 

Vu, T.P.T. (2025) The impacts of Gen AI in transforming social media marketing: a study of solopreneur social media operations.

Woolley, J. (2024) Generative AI and business: A review and research agenda. Oxford Research Encyclopedia of Business and Management. Available at: https://oxfordre.com/business/display/10.1093/acrefore/9780190224851.001.0001/acrefore-9780190224851-e-434 

Yadav, A.B. (2023) Gen AI-Driven Electronics: Innovations, Challenges and Future Prospects. International Congress on Models and Methods in Modern investigations (Poland), 2. doi:https://doi.org/10.5281/zenodo.8165255

Zhang, P. & Kamel Boulos, M.N. (2023) Generative AI in medicine and healthcare: promises, opportunities and challenges. Future Internet, 15(9), p.286.