Industry 4.0 introduced connectivity through IIoT-driven cyber-physical systems (CPS), enabling agile operations integrating physical assets with digital technologies, and ‘synchroperation’ is revolutionising the way in which manufacturing operations are managed. This shift synchronises human–machine–material interactions through agile and resilient operations driven by data analytics. It is supported by the Hyperconnected Physical Internet-enabled Smart Manufacturing System (HPISMS) and the Graduation Intelligent Manufacturing System (GiMS), which enhance visibility and traceability in manufacturing (Guo et al., 2021). There are two implementation progress measures for I4.0. One is INCIT’s Smart Industry Readiness Index (SIRI) benchmarking tool from the Singapore Development Board where assessors evaluate and prioritise advancing I4.0 with actionable plans (Winter, 2023). The other Acatech’s I4.0 Maturity Index evaluates six key trends: trustworthy data sources, industrial AI, operational migration to 5G connectivity, edge computing with cloud storage, team robotics and autonomous intralogistics systems (Kagermann & Wahlster, 2022).

The DTx gap is determined by recent global conflicts and disruptions such as coronavirus disease 2019 (COVID-19) highlighting limitations in resilience and sustainability within I4.0 frameworks and supply chain shortcomings (Borregan-Alvarado et al., 2021; Zheng et al., 2020). It encouraged the move to digitisation and automation with human intervention moved out of the loop. As a response, I5.0 emerged and addresses these shortcomings by integrating resilience frameworks with human-centricity and smart manufacturing principles, placing humans (‘back in the loop’) at the core of production processes. The focus is on worker welfare, sustainability and robust HMI central to the smart manufacturing process (Yang et al., 2024). Sarkar et al. (2024) highlight the International Standards Organization (ISO) White Paper (Johannsen, 2021) definition of smart manufacturing as improving operational performance through intelligent integration across cyber, physical and human spheres.

Many organisations have a hybrid relationship between the two paradigms (Golovianko et al., 2023). The Duggal et al. (2022) hybrid model blends the smart processes of both paradigms and sets the foundation for Industry 6.0 (I6.0). Here AI and robotics are taken to a level where human or users’ physical profiles are cloned as Pi-Mind clones or a digital twin of the user. These are designed to overcome the need to physically input product data to complete the manufacturing cycle; it is already integrated and cloned into the process. This becomes the differentiator for I6.0 where humans and machines learn collaboratively from each other (Duggal et al., 2022).

Resilience in Figure 1 is one of the most important enablers of I5.0 where Linnosmaa et al. (2021) suggest the introduction of a resilience framework comprising decision-making flowing from human operators assisted by decision support systems. In turn, this moves on to AI-driven autonomous control components with low level automation controls which can withstand internal or external negative disruptive influences and maintain its sustained operational performance which leadership must understand.

The future smart factory, according to Xu et al. (2021), incorporates cooperation between humans and machines where human-centricity, resilience and sustainability form core values. Yang et al. (2024) define HMI as integrating data processing transmission mechanisms with hardware and software collaboration. This enables optimising operator interaction efficiency, productivity and safety in HCSM. Combined with advanced technologies, this creates resilient and sustainable smart factories.

Figure 1 forms the basis of this study, integrating the three categories of HMI and four supporting category taxonomies which combine with IndAI, its empowering technologies and three cornerstones of I5.0 (Leng et al., 2024; Yang et al., 2024).

Figure 1 shows leadership how each category taxonomy and their empowering technologies are assessed against their contribution to the characteristics of HMI.

Cárdenas-Robledo et al. (2022) share that the IIoT connectivity enablement of technologies allows data exchange between people, systems and machines. The first taxonomy, data management, forms the basis for smart manufacturing enabling data integration, control and automation. Laroui et al. (2021) indicate that wireless transmissions are essential to the IIoT for reliability, security and timeless requirements, but cannot keep up with demand. Low latency (10 Gbps – 20 Gbps), high-band spectrum and fibre-enabled speed through 5G connectivity overcome this challenge.

Important aspects of data management and connectivity include data analytics with cloud and edge computing, together with machine and deep learning. Alouffi et al. (2021) advise that CC, a mature technology, employs flexible data management models with cybersecurity risks. Robust blockchain-based encryption mechanisms can effectively address intrusion and data leakage threats.

The Wang et al. (2022) opinion is that machine and deep learning algorithms are inseparable from the analysis of big data from manufacturing with correlation, prediction and regulation analysis characteristics. Edge computing, which possesses data closer to the source, mitigates latency concerns while supporting real-time industrial operations (Alouffi et al., 2021).

Machine learning algorithms have three learning categories according to Kotsiopoulos et al. (2021): supervised (given data with prediction labelling), unsupervised (discovering information from data), and reinforcement (given data and choosing actions for long-term reward). Unsupervised learning uses algorithms such as Bayesian networks and multiple linear regression. Supervised learning uses algorithms such as k-means and self-organising maps. Reinforcement learning uses algorithms such as PILCO and SMART.

Deep learning, although applied in manufacturing operations and processes, cannot guarantee industry safety and control requirements. Deep learning has applications in anomaly detection, image recognition, high dimensional data processing and natural language processing, relying on large data sets which become inaccurate when applied to smaller data sets. It is also associated with higher computational costs with higher hardware requirements (Kotsiopoulos et al., 2021).

Machine and deep learning have three key research development directions: task for quality and defects, technology algorithms for transforming industrial data to big data and industry-centred research on migration of new technologies from I4.0 to I5.0. The importance of deep and machine learning with industrial artificial intelligence (IndAI) as the brain of future smart factory scenarios will have key focus on human and environmental-related data areas of data analysis optimising the HMI experience (Kotsiopoulos et al., 2021).

Leadership must grasp how these support data management with the next taxonomy, a consideration of sensors in data detection.

Yang et al. (2024) agree with Kumar and Lee (2022) that I5.0 migration processes for smart manufacturing require five types of sensors:

In the third taxonomy, according to Yang et al. (2022), manufacturing has two types of data transmission mechanisms: the wired Industrial Ethernet and wireless sensor network (WSN). Wireless spectrums are either 5G or 6G transmission mechanisms (Zong et al., 2019). Wireless sensor network is better suited as an enabling technology for I5.0. It meets human-centric and sustainability requirements through 6G connectivity with increased data speed and coverage (Yang et al., 2024).

Three main issues are associated with HMI implementation: the allocation of tasks, workload allocation and trust (Yang et al., 2024). Kumar and Lee (2022) point out that smart factories, having increased technological complexity, face two areas of challenge. Firstly, worker mental load requires increased skill for advanced operational equipment. Secondly, with physical workload decreasing, data management demand increases with provision and collection of data from human-centric smart manufacture (HCSM) processes.

Janssen et al. (2019) highlight that a lack of trust in a system by workers leads to differing device and equipment usage, reducing efficiency and safety as complex systems increase operator confusion. Sustained system use will encourage workers to adapt overcoming misuse and disuse.

Additional collaboration and implementation challenges include:

For leadership, the goal of I5.0 is for humans and machines to work together without compromising workers’ mental health when working with HCSM data and models. The four category taxonomies examine challenges associated with three characteristics of HMI for HCSM implementation which have associated benefits and opportunities emerging.

New technologies driving the development of HMI include worker safety, health monitoring, and ergonomic workspace design. Aligned technologies include BCI, digital twins and the industrial metaverse (Yang et al., 2024).

Manufacturing execution system manages workshops across multiple functions, resource allocation and management, labour, process and quality management, document control, performance analysis, job scheduling, data collection and product unit scheduling (Yang et al., 2024). Shojaeinasab et al. (2022) indicate that the current MES systems available such as POMSNet Aquila, Siemens Opcenter, ABB MES, GE’s Digital Proficy MES and DELMIAworks as commercial systems generally lack advanced intelligent levels.

To enhance MES capabilities, recent research explores the integration of AI to improve efficiency, safety, quality and productivity. Additionally, digital twins are being developed to bridge MES with the physical environment, enabling real-time monitoring, review and data capture. Augmented reality is also proposed as a means of facilitating greater human–system interaction and collaboration within MES environments. These technological advancements for leadership are critical to realising the vision of I5.0 in HCSM and warrant further investigation, particularly regarding their alignment with sustainability goals and IndAI (Tao et al., 2024).

Morris et al. (2023), specify three grades of AI, artificial narrow intelligence (ANI), artificial super intelligence (ASI) and levels of artificial general intelligence (AGI). Industrial artificial intelligence can be broken into five levels (lower to higher): reactive, knowledgeable, learnable, autonomous and shareable. To determine which grade of AI a model belongs to simplifies the designing of appropriate models and algorithms with manufacturing digital upgrading is difficult. Cost-effective application and system reliability are critical to unlock the power of IndAI. Leadership must be aware of these IndAI levels and grades to ensure the optimum configuration is selected (Morris et al., 2023).

Figure 1 shows I5.0 has three considerations (Ghobakhloo et al., 2022): firstly, human-centricity where workers return to the centre of manufacturing acting as the main process decision-makers. Secondly, sustainable development through recycling, reusing, reducing waste and environmental impact with the industrial ecology forming good symbiotic development patterns. Thirdly, resilience, where brittleness is a system problem and not easy to correct as evidenced by the COVID-19 pandemic in global supply and value chains. Key resilience issues (Leng et al., 2024) in I5.0 include modularised manufacturing, flexible business processes, predictive controls, adaptable production capacity and a flexible strategic value chain. Such robust industrial systems essential to I5.0 require further investigation. Industry 5.0 fundamentals (sustainability, resilience and human-centricity) are symbiotically linked with the three IndAI opportunities (collaborative, self-learning and crowd intelligence) as shown in Figure 1 (Leng et al., 2024):

Kim et al. (2020) share that modelling agents can empower both humans and machines resulting in four significant areas of benefit: (1) collaborative decision-making through democratised access to AI models; (2) a new form of self-organising intelligence aligned with interoperability, data sovereignty, privacy, modularity and scalability; (3) increased governance supporting beneficial network behaviour and (4) IndAI in a more favourable decentralised position can attract users. Leadership must understand that designing appropriate models and algorithms that upgrade the manufacturing industry is of greater importance to unlocking the power of IndAI (Leng et al., 2024).

In Figure 2, Leng et al. (2024) show in detail components of a four-layer reference framework for IndAI applications consisting of the hardware infrastructure, core algorithm, computing engine and industrial application layers with their supporting and empowering technologies.

Industrial artificial intelligence implementors are advised to follow these four key principles. It is crucial for leadership to understand these, which leads to consideration of the contribution of empowering technologies for implementing IndAI.

The empowering and aligned technologies for implementing IndAI impact the four-layered architecture as shown on the right of Figure 2. These include cobots, secure-multiparty computation (SMPC), human-cyber-physical systems (HCPS), cyber-physical-social systems (CPSS), central processing units (CPUs), general processing units (GPUs) and AI with digital twins, and blockchain shared computing:

Human-cyber-physical systems control lifecycle processes, and CPSS bring personalised services for humans placing workers at the centre of the manufacturing system to interact with machines via intelligent HMIs (Leng et al., 2023). Digital twins are an enabler of HCPS through intelligent decision-making systems. Human-cyber-physical systems and CPSS have challenges with security and privacy using IndAI (Yilma et al., 2022):

Leadership must grasp the challenges associated with the contribution of the aligned empowering technologies in implementing IndAI. Future research requirements are covered in section ‘Limitations and future research directions’.

Van Erp et al. (2024) describe I5.0 as the integration of sustainability, resilience and human-centricity into industrial value creation. They propose a two-dimensional model incorporating design and operations referred to as the DesOps model for the DTx of I4.0 to I5.0 and HCSM having two criteria. Ecosystem and culture creation, which embodies four structural components of foresight, maturity assessment, objectives with key results and training. In turn, the first two components, foresight and maturity assessment, house system design, domain-specific design and continuous feedback, making up the design component of the model or Des. The last two components, objectives with key results and training, house continuous system integration, implementation and monitoring, making up the operations component of the model or Ops. The resultant DesOps model is fully integrated across all components and impacting all parameters of I5.0 sustainability, resilience and human-centricity (Van Erp et al., 2024).

In summary, leadership in manufacturing is advised to embrace and enhance all aspects of HMI, HCSM, HCI, I4.0 to I5.0 migration, worker back in the loop with worker safety, privacy and personal data protection, sensors, connectivity, data management and analytics, networks 5G or 6G and challenges associated with HMI implementation, future direction and trends. In addition, the impact of IndAI on I5.0 and HCSM, its architecture layers, grades and levels, characteristics and unlocking the power of IndAI with empowering technologies requires a thorough understanding with aspects of future IndAI research directions.

The original quantitative study targeted CEOs from a stratified sample of 2500 South African manufacturing businesses capable of DTx (Gaffley & Pelser, 2021). A structured questionnaire evaluated via SPSS software measured variables including demographics, data handling practices, leadership digital capability, strategy development frequency, and human capital skills development. Statistical methods included descriptive statistics, one-sample t-tests, chi-square tests, Pearson’s correlation, Spearman’s correlation, ANOVA, factor analysis and binomial tests (Gravetter et al., 2020). Gaffley and Pelser (2021) in the original quantitative research established that leadership in manufacturing had to understand the technological advances in the digital era and how these shape the digital future of their organisations. The research showed leadership had experience but lacked technical dexterity, with the younger generations technically adept but lacking experience. The cost of DTx projects requires C-suite agreement for budget compilation, spending and monitoring (Gaffley & Pelser, 2021).

Sarkar et al. (2024) conducted qualitative research into the critical success factors (CSFs) for smart manufacturing transition to I5.0, developing an interrelated CSF model from their findings. Their research was conducted across a sample of seventeen experienced senior manufacturing executives (Table 1) ranging from 7 to 22 years drawn from heavy manufacturing, technological transformation specialists, quality control, machinery parts and components and supply chain. Their research findings, through application of the Bayesian best-worst method (B-BWM), crystallised fourteen CSFs pertinent to I5.0 for a sustainable future (Table 3).

The study’s research objectives are directed at leadership to enhance their understanding of the technological advances in DTx gap analysis with migration of I4.0 to I5.0:

Author contributions in section ‘Human–machine interaction’ to ‘section Manufacturing execution systems and industry 5.0’, Table 2, provide 14 CSFs supporting future developments in data management, sensors, hardware, WSN networks with 5G and 6G network connectivity for data, HMI and HCSM.

The qualitative research in section ‘Research methods and design’, Sarkar et al. (2024) identifies 14 CSFs encompassing economic, social and environmental requirements of resilience and sustainability for migration to I5.0. These are grouped into human and technological categories (Table 3).

From section ‘Levels and grades of industrial artificial intelligence’ to section ‘Manufacturing and sustainability for industry 5.0’ on IndAI implementation, an additional nine CSFs are extracted (Table 4).

These 37 CSFs pertinent to migration to I5.0 require inclusion, enabling a more robust I5.0 DTx strategy formulation process detailed in section ‘A digital transformation strategy model for leadership adaptation to industry 5.0 smart manufacturing implementation’.

The first prerequisite is the appointment of a cross-functional DTx project team, drawn from mechatronics, engineering, IT, supply chain and services (human resources and finance).

STEP 1: The first team priority is to determine the baseline DTx gap. The outcome is benchmarked against a set industry standard or internal business objective. The original research has a DTx conversion of 47.7% in the South African manufacturing sector (Gaffley, 2022; Gaffley & Pelser, 2021).

Alternatively, four data parameters (analytics, storage, management and data decision-making) and four technological parameters (IIoT connectivity, digital twin with CPS, cybersecurity with blockchain and new future technologies) are evaluated on a scale 1 (least progress) to 10 (most relevance) for each parameter. These coordinates are linked across eight axes with scores aligned and joined in a spider web configuration (Kayali et al., 2023). The resultant set of lines visually has their spacings as before (blue) and after (red) depicting the broad DTx gap as STEP 1. The gap is then evaluated against 37 CSFs from Table 2, Table 3 and Table 4 coded for inclusion in Steps 2 to 6 in Figure 3:

STEP 2 – Leadership Digital Strategy (L): Considers and identifies visionary external uncontrollable variables impacting DTx strategy development for leadership evaluation, as these are factors over which the business has little or no control.

STEP 3 – Business Strategy Alignment (B): Defines the internal controllable dependent DTx variables relevant to goals which leadership has control over.

STEP 4 – Technology Implementation (T): Considers all technologies that expand and prioritise digital assets analysis outcomes.

STEP 5 – Data Management (D): Considers all aspects of data management covered in the literature review.

STEP 6 – Human Capital Development (H): Considers all aspects of human ‘back in the loop’ for HMI, BCI, HCSM and skills requirements.

STEP 7 – Performance monitoring review with cross-functional teams assigned clearly defined roles and responsibilities with aligned KPIs.

Review the process every 3–6 six months as DTx is a moving target with frequent technology, data innovations and improvements.

The structured approach enables leadership in manufacturing to effectively bridge gaps leveraging emerging I5.0 opportunities inherent within industry transitions.

Future research directions aligned with migration from I4.0 to I5.0 implementing IndAI have two sets of future challenges, social and technological barriers, as depicted in Figure 4 (Leng et al., 2024).

Figure 4 shows social and technological barriers with social barriers having two ethical considerations and three supporting sets of guidelines for HMI in I5.0 (Huang et al., 2023). Similarly, the technological barriers are characterised by the need for production control over various professions, domains and functions in four key areas of operation (Leng et al., 2024). Leng et al. (2024) advise that a systematic understanding of IndAI technologies is recommended for leadership to evaluate the correct technology application to unlock IndAI as an integral part of the DTx processes they lead.

Future research directions would include the aspect of sustainability and its impact on HCSM in I5.0, migration of I5.0 to I6.0, 6G connectivity, the industrial metaverse, digital twins, blockchain for personal data, operational security and supply chain, advanced cyber optic systems and IndAI advances in smart manufacturing. Authors in the literature review highlight that some of these technologies are in incubation and experimental stages with outcomes yet to be defined. These need constant evaluation before progressing their implementation.