Exploring AI-Driven Transformation in Management Paradigms for Recurrent Safety Hazards in University Laboratories

Abstract

The persistence of recurrent safety noncompliance (RSN) in university laboratories presents a critical challenge to laboratory safety risk management. This paper deconstructs RSN by conducting an in-depth analysis of potential safety risks, their underlying causes, and management obstacles. The research reveals that the phenomenon of RSN is fundamentally the result of the combined effects of complex human factor risks and outdated management methods. At the human factor level, cognitive biases regarding experimental safety risks and negative resistance lead to “habitual violations” of safety regulations. At the management level, routine laboratory safety inspections, requirements for rectifying safety hazards, and commonly adopted punitive measures have proven insufficient to prevent RSN. To address this issue, this study proposes actively leveraging the advantages of artificial intelligence (AI) in dynamic perception and proactive interventions. It advocates for the deep integration of AI technologies into the transformation of the management paradigm for RSN in university laboratories. Furthermore, this study preliminarily explores the application prospects, applicable principles, and scope of application of AI technologies in this context, providing an important reference for enhancing the systematic management of RSN in university laboratories.

1. Introduction

University laboratory safety is an important prerequisite for maintaining campus stability, ensuring the health of teachers and students and promoting technological innovation [1]. To achieve this goal, the laboratory safety management department typically needs to organize safety management personnel and experts to regularly or occasionally conduct safety inspections in laboratories, identify potential safety risks and hazards, and urge laboratories to fulfill their responsibilities to rectify these hazards. Whether these hazards can be permanently and effectively eliminated is key to maintaining the safety status of the laboratory, which directly impacts the overall safety environment [2].

However, the current situation is not optimistic. During the re-inspection of safety conditions in the same laboratory, it is often found that certain types of safety hazards recur, leading to the phenomenon of recurrent safety noncompliance (RSN)—a phenomenon in which specific safety violations resurface cyclically despite repeated interventions. In this paper, the RSN of laboratory safety refers to a repeated safety hazard that reoccurs in subsequent inspections after being identified and rectified previously, with the cumulative number of occurrences being greater than or equal to three. In this case, this safety hazard can be defined as an RSN-type safety hazard. For example, the laboratory safety management department requires experimenters not to place ordinary non-heat-resistant plastic products into a drying oven for a long time, especially not for overnight drying, in order to avoid the risk of fire. However, there are always some experimenters who violate the regulations, and even after repeated warnings, they still do not comply with this requirement, hoping that they will not encounter problems. In this case, such violations can be classified as an RSN-type safety hazard. The RSN phenomenon not only weakens the authority and effectiveness of laboratory safety inspections, but also greatly consumes the limited resources for university laboratory safety management. This results in the ineffective elimination of laboratory safety hazards and hinders the optimization and upgrading of laboratory safety management capabilities [3].

Therefore, managing the RSN phenomenon is an essential requirement for the effective management of laboratory safety in universities. Relying solely on traditional regular or occasional personnel safety inspections is inadequate to prevent the occurrence and spread of this phenomenon. Firstly, insufficient supervision by overburdened safety management departments hinders timely and effective corrective actions. Secondly, human error, whether intentional or unintentional, exacerbates the issue when safety protocols are neglected due to factors like time constraints and complacency [2,3]. Lastly, the inherent complexity of laboratory work makes it challenging to consistently maintain safety standards. Given the challenges posed by the RSN phenomenon in university laboratory safety, exploring innovative solutions becomes crucial. The potential application of artificial intelligence (AI) technology in laboratory safety management presents a significant opportunity for addressing the RSN phenomenon. Artificial intelligence, a multidisciplinary technology with machine learning as its core subset, has achieved tremendous success in the automotive and industrial manufacturing sectors [4]. In the field of laboratory safety, it also demonstrates great potential. For example, through intelligent cameras and various sensors, AI employs deep learning algorithms and artificial intelligence models for the real-time monitoring of the laboratory environment, recognition of unsafe behaviors in images, big data analysis, and risk assessment with early warnings [5,6,7]. Nevertheless, there are currently few specialized research studies aimed at addressing the RSN phenomenon in university laboratory safety, and discussions on the application of artificial intelligence in this regard are even rarer. Therefore, this paper attempts to deeply analyze the key reasons for the occurrence of the RSN phenomenon and explore how to utilize artificial intelligence technology to reduce its frequency and impact. Consequently, it aims to enhance the precise management level and efficiency of university laboratory safety and ensure the safety of campus laboratories.

2. Risks, Causes, and Management Challenges of RSN

2.1. Safety Risks Caused by RSN

Typically, university laboratory safety management departments develop their own laboratory safety inspection checklists (or standards, guidelines, etc.) or base them on checklists issued by national educational authorities, professional societies, and relevant laboratory safety organizations [1,2,3]. These checklists specify common items and key points for laboratory safety inspections, providing the sources and risk rationales for each inspection point while offering comprehensive and well-established content. Consequently, the safety risk assessment criteria of RSN are based on these types of laboratory safety inspection checklists.

Given the significance of laboratory safety inspection checklists in identifying and mitigating hazards, it is crucial to understand how recurrent safety noncompliance (RSN) can undermine these efforts and generate systemic safety risks. RSN creates systemic safety risks through three primary pathways:

(1)

It may cause persistent noncompliance, which prevents hazard rectification and fosters entrenched unsafe practices among stakeholders;

(2)

If the RSN phenomenon is not effectively managed, it will escalate tensions between laboratory units and safety regulators, delaying critical interventions;

(3)

RSN will also bring overlapping unresolved hazards that transform safety inspections into mechanical exercises, exponentially increasing the probability of accidents.

2.2. Individual Factors of RSN Occurrence

To understand the systematic risks caused by RSN, we explore the individual factors contributing to its occurrence. Human factors dominate laboratory safety incidents, with psychological states directly mediating unsafe behaviors [8]. Our analysis identifies three cognitive patterns driving RSN among laboratory personnel:

(1)

Viewing hazard rectification as personal criticism or resource encroachment, leading to superficial compliance without systemic resolution;

(2)

Considering safety protocols as obstacles to research/teaching efficiency, fostering resistance to evidence-based improvements;

(3)

Marginalizing safety measures as threats to academic freedom, particularly manifest in senior researchers who prioritize output over occupational health.

While these individual cognitive patterns significantly influence the persistence of RSN, it is essential to recognize that institutional structures also play a crucial role in either mitigating or aggravating these behaviors.

2.3. Organizational Architecture Contributors

Beyond individual factors, institutional structures create systemic RSN-permissive environments through three stakeholder groups:

(1)

Laboratory Personnel: Knowledge gaps, habitual noncompliance, and low accountability lead to RSN;

(2)

Safety Leadership: Conflicts in prioritization between short-term research outputs and long-term safety investments perpetuate management inertia;

(3)

Administrative Units: Inconsistent enforcement of reward/punishment mechanisms and underdeveloped safety cultures undermine systemic accountability.

Effective RSN mitigation requires tripartite collaboration rather than disproportionately burdening laboratory personnel with safety compliance responsibilities. Having explored the individual cognitive patterns and organizational architecture contributors to RSN, it becomes evident that a comprehensive understanding of these factors is crucial for developing targeted intervention strategies. The following section delves into a case analysis of the RSN phenomenon in university laboratories, providing concrete examples and insights to further illustrate the complexities of this issue.

3. Case Analysis of the RSN in University Laboratories

The case analysis below helps intuitively understand RSN-type safety hazard characteristics in university laboratories. This analysis is based on research conducted from January 2022 to September 2024. During this period, the researchers conducted 19 safety hazard inspections on the laboratories at Center B of their university and performed a statistical analysis on the identified hazard data. Despite a total of 342 safety hazards being identified, a significant portion of them were RSN-type hazards that had been repeatedly inspected and corrected, rather than new safety hazards. Figure 1 visually categorizes and displays some of these RSN-type hazards:

(A) Laboratory personnel lacked appropriate personal protective equipment (PPE): Figure 1A documents the RSN phenomenon in which laboratory personnel conducted experiments without wearing necessary protective gear (such as lab coats and gloves), increasing their risk of exposure to hazardous substances and injury. This lack of compliance with PPE protocols, along with the recurring nature of the issue, indicates insufficient internalization of safety culture;

(B) Fume hood use violations: Figure 1B demonstrates the non-compliance behaviors of operators when using fume hoods. One common non-compliance behavior is that the adjustable glass sash of the fume hood is opened to a height much greater than the required 10~15 cm from the worktop, which may lead to the leakage of hazardous gases or particulate matter, affecting the laboratory air quality and personnel health. Another issue is that overcrowding the fume hood with excessive items compromises its actual ventilation effectiveness;

(C) Cylinders safety issues: Figure 1C illustrates the storage of gas cylinders not in compliance with safety regulations, with three cylinders not being effectively secured. This could lead to cylinder toppling accidents;

(D,E) Non-compliance with laboratory electrical safety standards: Figure 1D demonstrates laboratory outlets that are overloaded with a mess of cords. Similarly, Figure 1E shows that flammable materials like cardboard boxes and foam boxes should not be piled under the distribution box, as these conditions could lead to electrical fires or shock accidents [8];

(F) Non-compliant storage and management of chemical reagents: Figure 1F illustrates issues in the storage and management of chemical reagents, with the reagent cabinet lacking a chemical inventory register and hazardous chemicals being stored without proper locking mechanisms;

(G) Improper installation or obstruction of emergency showers and eyewash stations: Figure 1G shows violations in which obstacles are present within the 410 mm range below the sprinkler heads of emergency shower devices, which could impede their normal use during emergencies and delay accident response;

(H) Accumulation of items in laboratory common areas: Figure 1H illustrates the overaccumulation of items in laboratory common areas, such as cardboard boxes, unpacked equipment, solutions, reagents, lab coats, etc. This not only affects the cleanliness of the laboratory but also may obstruct emergency evacuation routes, increasing the risk of fire and other safety incidents.

(I) Untidy and disorganized laboratory environment: Figure 1I shows an overall untidy and disorganized laboratory environment, such as randomly placed reagent bottles and trash that has not been disposed of in a timely manner. This may lead to chemical leaks, slips, and other specific safety incidents such as reagent bottle breakage.

In conclusion, the subfigures in Figure 1 intuitively demonstrate the main categories and manifestations of the RSN phenomena in university laboratories. These hazards are often repeatedly identified during laboratory safety inspections without effective correction and are very common RSN-type hazards that greatly consume university laboratory safety management resources [8]. If these RSN-type hazards cannot be systematically and effectively addressed, their cumulative impact may be enormous, gradually eroding the laboratory’s safety defenses and turning preventable risks into statistical certainties, posing a significant challenge to laboratory safety maintenance. Consequently, adopting new techniques and methods is necessary to mitigate the occurrence of RSN phenomena and enhance the level of laboratory safety management, thereby ensuring the long-term safety of laboratories.

4. AI Technology Empowerment

It is evident that relying solely on conventional management approaches for safety hazard identification struggles to achieve continuous, around-the-clock oversight, leading to inefficiencies in addressing RSN, such as delayed responses, prolonged resolution cycles, and suboptimal outcomes. To overcome these challenges, it is essential to design and develop an AI-driven digital management framework aimed at mitigating RSN [5,6,7]. For instance, by focusing on persistently recurring high-risk behaviors, AI can utilize visual perception (e.g., camera systems), acoustic perception (e.g., audio sensors), and predictive decision-making algorithms to autonomously monitor the perpetrators [9,10]. This significantly enhances the efficiency and timeliness of hazard detection, thereby preventing the entrenchment of unsafe or erroneous habits among laboratory personnel.

By integrating AI, safety managers can systematically identify and address “low-level but stubbornly persistent” hazards, such as the following:

(1)

Personal protective equipment (PPE) non-compliance (e.g., failure to wear lab coats or wearing shorts/sandals);

(2)

Poor laboratory hygiene (e.g., cluttered workspaces and unsecured chemical storage cabinets);

(3)

Unsecured hazardous materials (e.g., unlocked chemical storage).

Furthermore, AI systems can cross-reference data across platforms to flag individuals with frequent high-risk violations, enabling targeted disciplinary or educational interventions. For example, this could include automatically linking offender identities to laboratory access control systems to impose temporary entry bans. However, AI-assisted management is not universally applicable. Two critical prerequisites must be met:

(1)

Clear hazard characteristics for accurate AI recognition;

(2)

Unambiguous attribution of responsibility without dispute.

As illustrated in

Table 1. Key points of safety hazards of RSN that can be addressed with management by AI.

Management Items	Key Management Points
Laboratory hygiene and daily management	Sleeping in laboratories Storing, preparing, or cooking food Eating, drinking, or smoking
Personal protective equipment (PPE)	Failing to wear lab coats or protective clothing Non-compliance with the use of required PPE such as safety goggles, protective gloves, safety helmets/protective headgear, respirators, or face shields Wearing long scarves, neckties, or loose clothing improperly when operating rotating machinery (e.g., lathes) Having unsecured long hair (not tied under work caps)
Electrical Safety	Unauthorized installation of electrical wires/cables or improper connection of electrical wires/cables Daisy-chaining multiple power strips Power strips placed directly on floors Obstructed access to electrical panels Storage of flammable/explosive materials (e.g., gas cylinders and waste tanks) near electrical panels/outlets
Heating Equipment Management	Placing heating devices on flammable surfaces (e.g., wooden tables) Storing flammable and explosive chemicals, gas cylinders, or debris near heating equipment Failing to maintain safe distances from electrical panels, outlets, or power strips Abandoning ovens/heaters without disconnecting power, verifying cooling to safe temperatures or unattended operation of open-flame equipment (e.g., resistance furnaces) Lack of monitoring during high-temperature experiments
Hazardous Chemical Storage	Chemical storage cabinets being unlocked
Water Supply and Drainage Systems	Pipes leaking or faucets dripping

, certain hazard types that are common in routine inspections, predominant in RSN cases, and meet the above criteria are well-suited for AI-driven identification and management.

Implementing AI-powered, closed-loop management for RSN aims to significantly enhance efficiency. This allows laboratory safety personnel to reallocate efforts toward safety education, cultural initiatives, and project-specific risk assessments, thereby addressing latent, high-severity hazards more effectively. Crucially, integrating laboratory video surveillance systems enables automated detection, real-time alerts, corrective actions, and the closed-loop management of repeat violations. This approach represents a critical pathway toward achieving high-efficiency, low-cost precision management in laboratory safety.

While these theoretical advantages are compelling, their practical validation requires systematic evaluation. This study explores the application of AI management in RSN mitigation from a methodological perspective, as AI has not yet been practically deployed in this study. Here, several aspects that need attention when evaluating the effectiveness of AI management on RSN are briefly summarized: firstly, one should compile detailed RSN event records based on historical laboratory safety inspection data; secondly, one should create simulated scenarios and train AI models to test effective intelligent models and algorithms for reducing RSN occurrences; then, one should pilot AI in real laboratories to accumulate RSN management data; finally, after a period of operation, one should compare the RSN occurrence rate and rectification efficiency before and after the implementation of AI to assess its effectiveness while also considering technical reliability, costs, and the acceptance of laboratory personnel. AI has obvious advantages and great potential in RSN management.

5. Conclusions and Prospects

Identifying and rectifying lab safety hazards are crucial for managing risks on campuses. However, the phenomenon of recurrent safety noncompliance (RSN) poses a significant challenge to the safety of university laboratories. This study conducted an in-depth investigation into the risks, causes, and challenges of RSN, revealing that RSN creates a self-reinforcing cycle in which unresolved safety issues exacerbate systemic risks.

The primary reasons for RSN are the negative habitual thinking, adverse psychology, and repeated rule-breaking behavior of laboratory personnel, coupled with the management organizational structure’s difficulty in adapting to RSN management. To mitigate this situation, it is suggested to use AI to help with RSN management. This includes using special computer programs to study past mistakes, cameras to monitor if people are wearing the right safety gear, and smart computer systems to determine the best way to deal with risks. This new way of managing safety using AI will make it easier to spot RSN and lower the costs of checking for compliance.

Considering the potential of AI in RSN management as mentioned above, it is also important to be aware of its potential limitations. The potential limitations of AI in RSN management may include the impact of data quality on model accuracy, algorithmic bias and accuracy issues, and privacy and ethics concerns. Meanwhile, there will also be challenges related to deployment costs, institutional adjustments at the management level, and responsibility allocation. These issues need to be gradually addressed in future research and applications.