The impact of AI implementation in higher education on educational process future: A systematic review

Articial intelligence (AI) has been playing a vital role in all life domains. A striking example is AI effective revolution in health and educational services during the COVID-19 pandemic. Therefore, this systematic literature review investigates how AI impacts higher education (HE) by focusing on its impact on education quality, the learning and teaching process, assessments, and future careers. This review uses a systematic qualitative research method. The data is collected in a systematic review of academic articles on AI impact on HE from 1900 to 2021 from the Web of Science, Scopus, and ERIC. The process went through a systematic inclusion and exclusion procedure based on date, language, reported outcomes, setting and type of publications. Articles selected were screened via Rayyan Software and coded using excel based on the following themes: education quality, learning and teaching, assessments, future careers, and ethics. The total number of articles included is 56. The results vindicate that AI plays an ecient role in providing better education quality services, practical learning/teaching, and assessments approach for a better future career. Likewise, AI impacts future employment, which entails that HE institutions should incorporate more AI to have better graduates that meet the future market requirements. However, studies on AI impact assessments, ethics and future careers are limited and require further investigation.


Introduction
We now live in an arti cial intelligence era (AI). The impact of arti cial intelligence (AI) on education and higher education is a hot issue of discussion and experimentation. AI is affecting education/higher education, which requires a concise description of skills universities must educate students to prepare them for an AI future of work (Grace Ufuk, 2020). Arti cial intelligence has resulted in novel teaching and learning solutions that have been tested in various settings. Apart from its impact on education, AI signi cantly impacts labour markets, industrial services, agricultural processes, values chains, and the workplace (Kelly, 2021). AI in playing adopting virtual intelligence by most educational systems worldwide and basically by higher institutions during the pandemic (Senel & Can, 2021).
Current studies indicate that AI is vital in education quality. Equally, Khare and Stewart (2018) believe that AI positively impacts students from a life-cycle perspective using chatbots to support learners and provide better services. AI is also used in auto-grading, constructive feedback, and academic advising (Khare & Stewart, 2018). A Russian study highlighted educational services perspectives diversifying AI, the fourth revolution, and the social consequences of digitalisation markets. Results revealed that AI stimulates academic and teaching staff reduction. Meanwhile, the economy's digitalisation leads to new education services as entrepreneurship, nancially independent institutions, and AI trained staff (Bogoviz et al., 2019).
An analogous Ecuadorian study on how universities gather information about students and predict academic outcomes for proactive curricula, students' attention, and resources management. This study was based on neural networks and AI to parametrise perception. The results revealed the signi cance of AI Likewise, Alyahyan and Düştegör (2020) consider students success as institutions' performance matrices. Therefore, they investigated AI's role in predicting at-risk students and taking preventive measures for better performance. Results revealed AI e ciency in mining data, addressing issues and students' needs.
Consequently, AI-supported academicians and institutions decide to bring all potentials of success (Alyahyan & Düştegör, 2020). Equally, Vinichenko, Melnichuk and Karácsony, (2020) studied the most e cient technologies to bridge employee motivation and institutions' incentives using motivational AI. The study investigated academic motives and the university stimulating effect connection. Findings revealed a discrepancy between motivation and stimulation, which has impacted innovation ful lment. These gaps required applying innovative systems based on AI to meet the digital economy's requirements of the 21st century. The use of AI improved the staff creative competitiveness and impacted citation index and academic reputation. AI brought advanced solutions to the emerging problems by motivating staff, reducing the imbalance in staff motives and university incentives, and increasing staff publication and grants (Vinichenko et al., 2020).
Finally, AI and robotics are having a long-term impact on HE. These impacts are technical and pedagogical as well as social (Cox, 2021 The taxonomy expanded upon D'Mello and Graesser (2015) dichotomy of emotion-aware systems to guide a system that can foster positive emotions along the learning process via different approaches, design features and data sources. Comparably, Kaplan and Haenlein (2019) de ne AI as the system's ability to interpret, learn and apply data to achieve de ned goals and tasks via a exible application. The study analysed how AI differs from the internet of things and big data. Thus, it suggested looking at AI via evolutionary stages such as narrow and general AI or focusing on other systems such as human-inspired and humanised scholars. Results presented a framework that helps an organisation consider AI's internal and external implications in three labels C-model: Consonantly, Tashfeen (2019) explored how policymakers see education future within the continuing technology disruption. Thus, two learning alternatives were developed a vignette approach of the emerging technologies and space scenarios framework. Findings revealed that future scenarios involving cooperative styles like human machines cooperation and active virtual learning produce more desirable bene ts for education stakeholders. Implementing AI, 5G and automation will customise HE delivery and work landscape (Tashfeen, 2019).
Online learning increases demand, arti cially intelligent teaching, machine learning, and teaching assistants experimented with AI teaching assistance in the USA without knowing how learners perceive AI assistants. Findings indicated AI teaching assistant e ciency, communication ease, and AI teaching assistant's eventual adoption. The researchers recommend using teacher assistants and further investigations to understand better the nuances of AI teaching assistants' learning and teaching (Kim, Merrill, Xu, & Sellnow, 2020).
Moreover, Xiao and Yi (2020) investigated AI's e ciency in HE teaching and education's remarkable regularity of individual subjects. The study suggested personalised education that considers learners' desires and societal development needs. Evenly a rm the ine ciency of traditional teaching methods in ful lling individualised learning. Therefore, AI is the best alternative to achieving personalised learning via data analysis and modelling methods. Thus, AI predicts and tracks students' performance based on individualised training (Xiao & Yi, 2020).
GNS, Sanjinis and Nardo (2012) developed AI fuzzy logic assessment methods and competencies using Bolivia's subjective and objective tutor and tutee features. Findings indicate the method's usefulness at all education levels (GNS et al., 2012). Comparably, Camps et al. (2016) argue that HE activities humanlike system applications of knowledge to analyse academic credits and validate them for students from other institutions. Thus, an AI system was created to validate higher institutions' academic credits. The approach followed elicitation, modelling and construction of knowledge base. The ndings indicate that the system achieved the goal and acted as an e cient tool validating academic credits analysis to a level of 89.4% (Campos et al., 2016).
Likewise, Radović, Petrović and Tošić (2020) highlighted the requirements of an e cient curriculum along with an improved and renewed assessment method.
The Comprehensive Integrative Puzzle (CIP) is one of the promising emerging practices. However, it requires high efforts, a team of experts, a native representative of the English language, and a time-consuming approach that presents an automatic generation of CIP assessments questions. Findings indicate that the adoption of ontological knowledge representation enables multilingual education domains. Medical education automation is one of the most challenging elds in the CIP. However, automation promises innovative teaching methods in all educational domains (Radović et al., 2020).
Correspondingly, sustainable AI requires ethical governance in increasing worldwide demand. China has enhanced AI technologies and published AI ethical guidelines and principles to bene t societies, companies, states, and research organisations by stressing: privacy, security, safety, reliability, accountability, transparency, and fairness. Equally, the Economic countries released AI ethical guidelines and governance regulations. In 2018, the EU announced the General Data Protection Regulation (GDPR). Equally, the White House issued AI executive order to maintain American leadership. The National Institute of Standards and Technology developed technical regulations for reliable, robust and trustworthy AI systems. Equivalently, the United Nations promised to promote AI ethics and called for an AI conference in 2019 stressing AI links with human values. Finally, Google, Amazon, Microsoft, Alibaba, Baidu and Tencent got involved in AI ethics and governance (Bozkurt et al., 2021).
AI is vital in knowledge management, learning, teaching and skills development. Thus, AI is needed along with human skills as it is a requirement of the future workforce. Automation and the gig economy radically change job descriptions and work methods, making human and AI skills, adaptability, and resourcefulness key to success (Taneri, 2020). Gong et al. (2019) highlighted the potential of AI in transforming radiology and clinical practices. The study targeted Canadian medical students and their perception of AI's in uence on their radiology majors' performance. Findings revealed that most 67.7% disagree that AI would replace radiologists, in contrast with a minority of (29.3%) who agreed that AI would replace radiologists in the future. Additionally, 48.6% were anxious considering AI in radiology. The radiology community was interested in expert opinions on AI and suggested that medical students educate about AI's possible impact on radiology (Gong et al., 2019a).
Likewise, LIorente (2020) stresses the various revolutions from engine steam to electricity and assembly lines, moving to the computer in 1960 and the fourth revolution. The fourth revolution involves AI, digital technology, globalisation and hyperconnectivity (G5). This revolution requires radical lifestyle changes and the workplace. The social labour situation faces signi cant di culties such as increasing unemployment, the ageing society, the emergence of new leading sectors in the job market, exponential technology development, the excessive use of robotisation and globalisation of the market. Consequently, radical changes require new educational systems, companies and work organisations. Results indicate automation will increase productivity, devalue salaries and eradicate tasks. E-workers will have different ambitions than the current ones. The concept of jobs will be replaced by career professionalism. The big corporate economy will replace the current one (Llorente, 2020). Finally, this systematic review investigates how arti cial intelligence affects higher education and studies the impacts of AI on education quality, the learning and teaching process and future careers.

Methodology
This review is a rigorous approach synthesising and amalgamating the most recent academic papers on AI's impact on HE since 121 years. The research question was formulated based on PICOT (patient/population; intervention/indicator, compare/control; outcome; and time/type of study or question) focusing on prognosis/prediction question: Does AI implementation in higher education impact the educational process in the future? The nal search equation was de ned using the Boolean connector "AND", and the keyword combinations were made, ensuring one keyword for each search category (Arti cial intelligence, impact, higher education). The review followed the PRISMA checklist for a systematic review, ensuring transparency and entirely reporting metanalysis and systematic review (Liberati et al., 2009). The review also used a systematic qualitative method to systematically search for evidence from primary qualitative studies and draw the ndings (Seers, 2015).
The preliminary process started with titles processing. Suppose the title is related to the scope of AI and HE. In that case, the article is selected and exported to Rayyan Software, including author, year, title and abstract for further screening. Data collected is secondary via desk research focusing only on academic articles relevant to the abovementioned topic and processed via the reference management App Mendeley. The limited researcher investigation to the Web of Science, Scopus, and ERIC across various disciplines ensures reliability and validity. The researcher created a research protocol made of ve features: (1) date, (2) language, (3) reported outcomes, (4) setting and (5) type of publication. The protocol used transparent criteria of inclusion and exclusion to shortlist articles researched. The researcher screened articles via Rayyan Software. The protocol was to minimise bias, ensure transparency, replicability, and as an indicator of feasibility (Choi et al., 2019).

Eligibility assessment
To ensure the adequacy of screening, the researcher used Rayyan Software, a free mobile and web app for systematic literature review provided by the University of Qatar. A total of 509 articles was found in the rst stage and exported via Endnote for screening following the designed inclusion/exclusion criteria. Therefore, the total number of included articles is 56 from Scopus, Web of Science and ERIC. The researcher thematically coded the articles via Excel.
Consequently, four major areas were highlighted, education quality, learning and teaching, assessment, future career, and AI ethics in higher education. (For further details, see gure 1)

Inclusion and exclusion criteria
The article's inclusion used the following criteria: 1900 to 2021, language limited to English only, and publication types were limited to academic articles. The setting was only linked to higher education institutions and AI. The review also was limited to only consistent, appropriate outcomes (The University of Melbourne, 2021). Overall, the study's scope focused on Human Science and disciplines like Computer Science. The exclusion criteria were as follows: exclude sources not written between 1900 and 2021, sources written in other languages than English, sources with inconsistent outcomes, sources not linked to higher education.

Findings
The nal selection included 56 scienti c articles that provided evidence regarding the educational quality, Learning and teaching, and AI ethics in higher education and Future Careers in Higher education. Table 1 summarises the information on the articles: reference, country, sample, research design, Key ndings. Findings address four crucial topics as summarised in the above table (1) Impact on the Education quality, (2) Impact on the learning and teaching process, and (3) Impact on assessments, and (4) Impact on ethics and future careers in higher education. The following part is a deep analysis of the signi cant results found.

Impact on the Education quality
Findings indicate that AI transforms education quality: AI helps learners communicate better and connect to the world (Breaux, 2017). . Likewise, AI processes structured and unstructured data to reduce management workload and speeds decision-making (Bojorque & Pesántez-Avilés, 2020). Equally, AI enhances problem-solving, creativity, time management and communication (Korepin et al., 2020). AI reinforces strategic planning and effective learning and teaching. Therefore, AI provides speedy, accessible data and intelligent analysis that leads to proper intervention at the right time (Prinsloo, 2020). AI boosts cognitive abilities, learning adaptability and timely decision making. AI enables instructors to take multiple actions simultaneously: grading, giving feedback and detecting at-risk students . Constantly, AI helps predict-risk students and e ciently perform data mining (Yahyan and Düştegör, 2020). Therefore, (Fayoumi and

Impact on the learning and teaching process
Results revealed a vital need for AI in teaching and learning (Williams, 1992

Impact on Ethics and Future Careers in higher education
Results reveal a crucial need for moral systems. Therefore, e cient technical, theoretical and psychological supplements must reinforce AI with spirit and conscience (Pana, 2006). Additionally, AI must be guided by ethics and principles to ensure that it is bene cial for societies, companies, institutions and humans worldwide . Concerning future careers, ndings indicate that AI and robotics will impact some professions, such as librarianship and radiology (Yildirim and Yildiz, 2018; Gong et al., 2018). Besides, results stressed that AI helps predict learners' future careers (García-Peñalvo et al., 2018).
Finally, it is needed to explore the rapid transformation of the quality and quantity of work socially, geographically, and at a governmental level (Clifton et al., 2020).

Conclusions
This literature review focused on the in uence of arti cial intelligence on higher education. The impact of arti cial intelligence (AI) on education and higher education is becoming a burning issue worth experimentation as AI is affecting life and education/higher education. The literature reviewed has covered four major areas: the impact of AI on educational quality, learning and teaching process, assessment, and ethics and future careers. The existing literature highlights the importance of AI in higher education and insists on implementing AI in the educational system and training academic staff and learners with AI for better educational systems and a better future. The signi cant gaps found are that studies conducted are about AI and higher education in general and do not focus on particular areas in education such as learning, teaching, assessment, and quality to assist the learner and tutor more. Inclusion and exclusion process Note: the researcher designed the gure