Phase 1: questionnaire developmentInterviews Participant characteristics

A total of 12 individuals participated (7 women, 58.3%; and 5 men, 41.7%) with ages ranging from 19 to 35 years (M 24.25, SD 4.37). The majority were students in the field of psychology (n = 6), while the rest were enrolled in diverse academic programmes (e.g. engineering [n = 2], criminology [n = 1], history of art [n = 1], pedagogy [n = 1] and primary education [n = 1]).

Emerging themes and sub-themes

Eight themes emerged from the analysis of interviews. Table 1 presents these themes along with their sub-themes accompanied by illustrative quotes.

Table 1 Emerging themes and subthemes from the interview data, with representative participant quotes

Eight sub-themes for resources for managing mental health were identified, of which talking with trusted individuals (n = 8) and engaging in physical activity (n = 5) were the most commonly mentioned. As for the theme usage of ChatGPT, all participants (n = 12) had used it at least once, primarily for academic, informational or organisational purposes. The usage frequency ranged from occasional to daily. Of these, five participants (41.7%) reported using ChatGPT specifically to support their mental health. A total of nine sub-themes emerged for the potential purposes of their usage, which was another theme identified. Their usage was primarily oriented towards seeking strategies (n = 6), gaining perspective (n = 5), and feeling understood (n = 4), particularly in relation to interpersonal problems (n = 4) and personal concerns (n = 3), which were the most frequently mentioned sub-themes for the theme potential thematic area. The remaining seven participants (58.3%) indicated that they had not used ChatGPT for mental health purposes, citing a preference for human interaction or concerns about data privacy.

The themes facilitators and barriers were identified with 11 and eight sub-themes each, respectively. The most frequently mentioned facilitators included immediacy (n = 7), access to a large volume of information (n = 5) and constant availability (n = 4). The most frequently mentioned barriers were the impersonal nature of the tool (n = 10) and its limited ability to be personalised (n = 6).

The Supplementary table presents the questionnaire used in phase 2 and indicates the items that were developed on the basis of information derived from the interviews.

Phase 2: cross-sectional studySociodemographic characteristics of the sample

The final sample consisted of 500 participants with a mean age of 41.15 years (SD 17.7; range 16–90). The majority of the participants (68.6%, n = 343) identified as women. Most participants (58.8%, n = 294) were employed at the time of participation. The participants reported a wide range of educational and professional backgrounds (see Table 2 for further details).

Table 2 Demographic characteristics of the participantsUse of ChatGPT

A total of 74.6% of participants (n = 373) reported having used ChatGPT at least once. Among these users, the frequency of use varied as follows: 26.8% (n = 100) reported occasional use, 19.6% (n = 73) used it once or twice a week, 24.9% (n = 93) used it three to five times per week, 13.4% (n = 50) used it daily, and 15.3% (n = 57) reported using it multiple times per day.

Among those who had never used ChatGPT (n = 127), the reasons ordered by popularity included insufficient knowledge about the technology (41.7%, n = 53), a preference for solving problems independently or with help from trusted individuals (32.3%, n = 41), and a lack of interest in the tool (13.4%, n = 17).

Characterisation of ChatGPT as a mental health management tool

Eighty-six participants (17.2% of the total sample or 23.1% of ChatGPT users) reported having used ChatGPT as a resource for managing their mental health. Among these users, the majority described using it infrequently. In fact, the number of participants reporting use decreased as frequency increased. The majority (58.1%, n = 50) used it once a month or less, while 15.1% (n = 13) used it 1–2 times per week, 11.6% (n = 10) used it 3–5 times per week, and smaller percentages reported using it every day (10.5%, n = 9) or several times a day (4.7%, n = 4).

In comparison to other commonly used strategies for mental health management, the use of ChatGPT was clearly minor. Participants more frequently relied on self-care and social support from friends or family members (Fig. 1).

Fig. 1

Percentage distribution of the frequency of use of different resources for managing mental health

Of the 86 participants who reported using ChatGPT to support their mental health, the most commonly addressed topics were stress, anxiety or general worries (64%, n = 55), followed by interpersonal problems (53.5%, n = 46) and personal growth (51.26%, n = 44) (Table 3).

Table 3 Mental health–related topics discussed with ChatGPT

Most participants reported using ChatGPT to obtain advice or strategies for emotional regulation (73.3%, n = 63), to gain a new perspective on their concerns (61.6%, n = 53) and to organise their thoughts (53.5%, n = 46). Less frequently reported purposes of use are detailed in Table 4.

Table 4 Purposes of using ChatGPT for mental health support

The perceived effectiveness of ChatGPT was mixed and almost split, with roughly half reporting moderate to high benefit (i.e. 36% (n = 31) stated that ChatGPT had helped them moderately, and 19.8% (n = 17) reported that it had helped them greatly), and the other half describing it as only occasionally or not at all helpful (i.e. 38.4% (n = 33) indicated it had helped them only occasionally, and 5.8% (n = 5) stated that it had not helped them at all).

Although the most frequent response was that ChatGPT had no impact on participants’ likelihood of seeking professional help (34.9%, n = 30), over half indicated some degree of influence: 30.2% (n = 26) reported a slight influence, another 29.1% (n = 25) a moderate influence and 5.8% (n = 5) a strong influence.

Perceived facilitators and barriers in the use of ChatGPT for mental health support

The most frequently reported facilitators, each endorsed by the majority of participants, included the speed of response (75.6%, n = 65), the general accessibility of ChatGPT (68.6%, n = 59), 24/7 availability (67.4%, n = 58) and ease of use (53.5%, n = 46). The most frequently reported barrier—also identified by the majority—was uncertainty regarding the reliability of the information provided (60.5%, n = 52). Overall, facilitators were more typically identified than barriers (Figs. 2 and 3).

Fig. 2

Facilitating factors for the use of ChatGPT in mental health management according to participants who have used it for that purpose (%)

Fig. 3

Factors perceived as barriers to the use of ChatGPT in mental health management according to participants who have used it for that purpose (%)

Use of mental health-specific chatbots

Only nine participants (2.4% of the total sample) reported having used a virtual assistant specifically designed for psychological or emotional support. Among these, Woebot and Wysa were each mentioned by one participant.

Use of ChatGPT by age group

A clear trend emerged indicating that younger participants were more likely to report having used ChatGPT (Fig. 4). Statistically significant differences were observed between age groups (χ2(5) = 154.6, p < 0.001). Similarly, with the exception of the group aged over 65, younger participants were more likely to report having used ChatGPT for mental health management (Fig. 4), with statistically significant differences again observed (χ2(5) = 21.4, p < 0.001).

Fig. 4

General and mental-health-related use of ChatGPT across age groups

Age-related differences in the perceived helpfulness of ChatGPT and its influence on professional help-seeking were assessed with Fisher’s exact test after excluding the two oldest age brackets (≥ 50 years) due to their small subgroup sizes (n < 5). No significant differences were detected (F = 4.5, p = 0.9, and F = 7.9, p = 0.5, respectively). See Figs. 5 and 6 for the distribution of responses across age groups.

Fig. 5

Perceived helpfulness of ChatGPT for mental-health support, by age group (%)

Fig. 6

Perceived influence of ChatGPT on professional help-seeking behaviour, by age group (%)

Usage facilitators and barriers by age group

Applying the same statistical procedure described in the previous section—Fisher’s exact test with prior exclusion of the two oldest age groups (n < 5)—no significant age-group differences were observed for any facilitator or barrier (all p > 0.05).