This study employs semi-structured interviews to meticulously capture the varying attitudes and behavioral responses of creative professionals as they interact with GenAI, with a focus on the stress of content creation competition. This method focuses on detailing people’s experiences and viewpoints and uncovering the significance they attribute to them in a particular context (Alon and Krtalić, 2024), such as subtle attitudes and responses experienced in approaching and adapting to GenAI.

From April 12 to 22, 2024, 28 initial participants were reached through interpersonal networks and online communities (e.g., Sina Weibo). To capture a wide spectrum of views, the interview pool was composed of 18 creative professionals, selected for their diversity in gender, occupation, experience, and usage. The interviews were conducted via Tencent Meetings and/or face-to-face. Among the participants, 53% were male and 47% were female, with ages ranging from 27 to 40 years and work experience spanning 2 to 10 years.

The interview primarily focused on our research questions, while participants were also encouraged to discuss topics beyond these, including but not limited to their perceptions of the challenges GenAI poses to their creations, as well as their perceptual changes related to GenAI use, its future development, and their professional growth. Each interview lasted approximately 1 hour and was transcribed into text for coding purposes.

The collected data were analysed using the qualitative analysis software MAXQDA, following a thematic analysis research method (Naeem et al., 2023). This technique included data familiarization, code selection, theme development, and conceptualization. During the coding phase, A two-stage coding scheme was employed: first, horizontal coding was used to examine each participant’s individual experiences; then, meanings clustering was conducted to identify and categorize overarching themes and sub-themes across all interviews (Alon and Krtalić, 2024). Two raters performed two independent rounds of coding on the raw textual material, with a third author participating in discussions at the end of each round to enhance coding consistency. The inter-coder reliability was assessed using Cohen’s Kappa, which was calculated to be 0.82 for all three coders, indicating a strong level of agreement in the coding process. The dimension stages, aggregate themes, and codes are presented in Figure 1.

The coding analysis shows that creative professionals’ engagement with GenAI involves initial exposure, adoption, and adaptive use. Figure 2 illustrates how participants’ perceived stress and value evolve at different stages.

For participants in the initial exposure phase, GenAI triggered significant stress, largely driven by occupational, technical, and organizational challenges. Occupational stress arises from GenAI lowering the creative threshold for professionals. P8 said, ‘Almost anyone can use text-to-image AI to create art, the creative jobs might disappear’. In addition, most participants reported that their creative occupation was classified as having high exposure to AI (Feltenet al., 2021) and was perceived as particularly vulnerable to AI replacement. While GenAI may be considered ‘good stress,’ its technology advantages—such as strong fusion capabilities, anthropomorphic interactions, personalized services, rapid responsiveness, and high-quality content—could still threaten professionals, as inefficient AI might not be seen as a stressor. For example, P9 said, ‘Why is AI advancing so rapidly? Because it truly learns from the data you give it, but people might not. It’s quite frightening!’ Organizational stress stems from cost reduction and efficiency-driven, results- oriented culture. A marketing creator mentioned, ‘For businesses, what once seemed like an impossible triangle (high quality, low cost, and short time) has become achievable with the advent of GenAI. We might not be able to reach it’ (P6).

During the adoption phase, participants conveyed that perceptions of self-value realization, such as remaining competitive, gaining work autonomy, human-in-the-loop (HITL) and perceived utility, helped alleviate the initial three challenging stresses associated with GenAI. Compared to not adopting GenAI, embracing it can lead to greater work autonomy and to enhanced competitiveness (P1, P4, P11, P17). The need for human involvement in bridging GenAI’s limitations strengthens the case for adopting GenAI, making it hard to move away from it under content creation stress. Similarly, the attraction of perceived utility seems to have weakened the stress on participants about content creative competition. As one participant noted, ‘It’s like an invisible assistant; I’m almost forgetting the career stress it caused’(P10).

When participants enter the active adaptation stage, adjusting and adapting to each other to GenAI will form their emotional value and subjectively enhance their sense of satisfaction, dependence, dominance, and technological identity, further influencing their initial perception of stress. For example, P15, a designer who has been using ChatGPT intermittently since its inception, shared: ‘The way GenAI and I can adapt to each other’s needs is satisfying and removes any need for unnecessary concern.’ The sense of dominance stems from the guidance and mastery of the GenAI tool. For example, one participant stated, ‘I can take control of GenAI and incorporate my ideas into the creation process to achieve my desired results’(P7). Technological identity refers to an individual’s identification with and sense of belonging to a particular technology. As one puts it,

AI, as a new quality productive force, has gradually reshaped the way I work in collaboration with it, and even if there is competitive stress, it is from other peers rather than AI (P11).

Additionally, some participants acknowledged a growing dependence on GenAI during the adaptation phase, with a high likelihood of path dependency, which exacerbates the pressure to use the technology. For instance, P4 remarked,

‘Before using ChatGPT, my work efficiency was 100%. After using it, it increased to 150% or even 200%. However, losing ChatGPT’s assistance may drop my efficiency to 50% instead of returning to 100%.’

Findings showed that creative professionals’ attitudes toward GenAI in their work varied with their perceived value experiences at different stages of interaction. Participants’ attitudes toward the use of GenAI may shift from initial worry to moderate concern, and eventually to less worry. These varying attitudes can trigger either positive or negative behavioral responses.

When participants were in the initial exposure stage, they often had a strong negative attitude toward GenAI. At this stage, due to the predominant sense of threat and work stress forced creators in compromise, participants mainly exhibit negative maladaptive responses, including adverse emotional reactions, resistance behaviours, and negative career choices and adjustments. For example, one programmer (P18) expressed concern that the sense of crisis in his profession caused by GenAI led him to either silently or vocally resist it, but this resistance proved ineffective. Participants also conveyed that when they are particularly concerned that AI-generated content might undermine the value and significance of their creativity, they may exhibit resistance by withdrawing, avoiding, or opposing the use of GenAI in their workflow. Job burnout, leaving, and changing jobs were also behavioral responses mentioned repeatedly by participants in this phase. For example, P1 said, ‘I used to be a programmer, and I quit because I was very aware of my career prospects. Sooner or later, my job would be replaced by AI.’

In the adoption stage, the stress of job replacement gradually transforms into a positive driving force for progress. Although creators adopt GenAI in their creative practices, the stress of competition can be activated at any time, causing them to fluctuate between an embracing attitude and concerns about their future careers. At this stage, some respondents take positive and constructive responses to navigate their challenges. They either adopt GenAI to assist with self-improvement and actively follow and recommend GenAI, or they begin to alleviate professional stress by continuously evaluating and critically reflecting on the use of GenAI. For example, the generation of mis/disinformation was identified by several respondents as one of the reasons they view GenAI with less tension. ‘I find that ChatGPT often uses strange but plausible logic to arrive at conclusions that only seem correct at first glance’(P2). Some participants (P2, P5, P12) cope with stress by over-discussing and sharing the negative experiences and flaws of GenAI in online communities, a behavior known as co-rumination (Hu et al., 2023).

The adoption of GenAI will further drive its exploratory and adaptive use (wael AL-khatib, 2023). At the adaptation stage, some participants exhibit adaptive behavioral responses. Reshaping creative roles is notable, such as data analysts need to transition into ‘AI model trainers and supervisors’(P12), program developers perceive themselves no longer as mere ‘code workers’ but as ‘managers, controllers, and learners’ of the technology (P1), Photojournalism will undergo a role transformation from ‘photographers to reporters to verifiers’(P5). Another response is to collaborate with multiple GAI agents. A game music producer shared how they use AI tools in their workflow,

for lyrics, I use ChatGPT and NetEase Tianyin; for composing, Suno generates accompaniments and melodies; for vocals, ACE helps with voice replacement. For a complex music video, I use Kaiber, and for animations, I turn to Runway and Pika (P9).

Finally, to protect themselves from job displacement as GenAI learns human behavior patterns, some creators deliberately mask their true needs during interactions with the technology. They use misrepresentative interactions to safeguard their job security and professional status. An advertising planner noted, ‘I avoid sharing my expertise with ChatGPT because it might surpass my skills and render my role obsolete’(P13).

Participants’ use of GenAI for their creative work depended on specific contexts individuals find themselves in, which encompassed three main factors: individual traits, organizational support, and social influence. These contexts may cause participants to linger in or move between different stages.

Findings on the individual traits of participants revealed that AI literacy and openness to innovation strongly influenced their stress, attitudes, and responses to GenAI. Those with higher AI literacy and a positive attitude towards innovation felt less competitive stress and were more excited than worried, demonstrating more positive adaptive coping behaviours. For example, P1, an illustrator, shared, ‘I do meet many co-workers who are scared and resistant to it, but I embrace change and continuously adapt GenAI in my work.’

If participants’ individual traits enable them to proactively learn, adopt, and grow with GenAI naturally, then the organization and leadership will need them to do so. For example, P19 recalls her experience with GenAI: ‘The company first encouraged independent use, then trained all staff on Midjourney and prompt skills, which I was initially reluctant to learn on my own.’ P14, in a leadership role, stated, ‘I have mandated my employees to use AI for creation...’

Social influence mentioned by participants, such as widespread support and promotion of GenAI in government policy and the media, also affects their engagement with GenAI. One mentioned, ‘Government departments and big companies are already using AI; it’s only a matter of time before it’s everywhere. Embracing GenAI is a smart move’(P16). Similarly, P17 observed, ‘The new AI rules show that everyone’s still backing the growth and use of GenAI.’