Figure 1
Sequential Mixed-Methods Design.
Figure 2
Gen AI Use of Swiss Public Sector Employees at Work (n = 702).
Table 1
Final Issues from the Brainstorming Round.
| ISSUES | SHORT DESCRIPTION |
|---|---|
| Reduce stigmatization | Aims to counter fears, misconceptions, and social biases of Gen AI |
| Demonstrate benefits | Highlights the tangible advantages of Gen AI in the public sector |
| Training | Focuses on structured programs that help understand how Gen AI works and how to use it effectively |
| Set an example | Encourages visible use and endorsement by public sector managers |
| Communication | Emphasizes transparent and proactive communication about the purpose, use, and limits of Gen AI |
| Clear guidelines | Covers the development and implementation of coherent operational, ethical, and legal frameworks |
| Coercion | Refers to regulatory or institutional measures that require the adoption of Gen AI |
| Ensure access | Seeks to guarantee equitable and inclusive access to Gen AI tools |
Table 2
Results of the Delphi Panel.
| ISSUES | ROUND 1 K = 8 | ROUND 2 K = 8 | ROUND 3 K = 8 | |||
|---|---|---|---|---|---|---|
| MEAN RANK | D2 | MEAN RANK | D2 | MEAN RANK | D2 | |
| Training | 2.94 | 2.28 | 2.58 | 3.12 | 2.24 | 3.22 |
| Demonstrate benefits | 2.94 | 2.28 | 2.39 | 3.79 | 2.39 | 2.70 |
| Ensure access | 2.97 | 2.19 | 2.42 | 3.67 | 2.18 | 3.44 |
| Set an example | 3.70 | 0.57 | 4.06 | 0.08 | 4.06 | 0.00 |
| Clear guidelines | 4.27 | 0.03 | 4.39 | 0.00 | 3.18 | 0.73 |
| Communication | 4.48 | 0.00 | 4.55 | 0.04 | 4.82 | 0.61 |
| Reduce stigmatization | 6.64 | 4.78 | 6.85 | 6.29 | 6.45 | 5.84 |
| Coercion | 7.67 | 10.34 | 7.48 | 9.88 | 6.97 | 8.60 |
| Totals | 35.61 | 22.48 | 34.73 | 26.88 | 32.27 | 25.25 |
| Grand Means | 4.45 | 4.34 | 4.04 | |||
| W | χ2 | W | χ2 | W | χ2 | |
| *p < .001 | 0.54 | 123.65* | 0.68 | 152.43* | 0.72 | 152.15* |
Table 3
Alignment of Delphi Results to the CFIR Domains and Constructs.
| ISSUES (DELPHI) | CFIR DOMAIN | CFIR CONSTRUCT | EXPLANATION FOR SWISS GOV-GPT |
|---|---|---|---|
| Ensure Access | Inner Setting | Available Resources | Ensure that the necessary infrastructure, tools, and support are available so that Gov-GPT can be accessed in an equitable manner. |
| Training | Individual Setting | Self-Efficacy | Provide targeted training to enable employees to use Gov-GPT in their daily work. |
| Demonstrate Benefits | Intervention Characteristics | Relative Advantage | Clearly communicate the added value, efficiency gains, and functionality of Gov-GPT compared with existing tools to strengthen its perceived usefulness. |
| Clear Guidelines | Outer Setting | External Policy & Incentives | Develop clear internal guidance for the appropriate use of Gov-GPT, aligned with applicable legal and regulatory frameworks. |
| Set an Example | Process | Engaging Opinion Leaders/Champions | Encourage leaders and key users to actively apply and promote Gov-GPT, fostering trust and normalization within the Swiss public administration. |
| Communication | Inner Setting | Networks and Communications | Strengthen internal communication and exchange on Gov-GPT implementation and maintain dialogue with other offices to ensure coordination. |
| Reduce Stigmatization | Individual Setting | Knowledge & Beliefs about the Intervention/Implementation Climate | Address concerns and misconceptions (e.g. job displacement, data security) transparently to promote trust and openness toward Gov-GPT. |
| Coercion | Inner Setting | Tension for Change/Incentives | Recognize that pressure for adoption may arise from political or regulatory requirements. Internally, ensure that implementation is guided by clear rationale. |
Figure 3
Positioning Delphi-derived Results within the CFIR Framework.
Table A1
Swiss Public Sector Employees Gen AI Usage at Work – Survey Results.
| FREQUENCY | MUNICIPALITY | CANTON | FEDERAL | TOTAL |
|---|---|---|---|---|
| Never | 190 (42.7%) | 58 (24.4%) | 4 (21.1%) | 252 (35.9%) |
| Rarely | 158 (35.5%) | 87 (36.6%) | 8 (42.1%) | 253 (36%) |
| Once a week | 46 (10.3%) | 37 (15.5%) | 4 (21.1%) | 87 (12.4%) |
| Several times a week | 43 (9.7%) | 47 (19.7%) | 3 (15.8%) | 93 (13.3%) |
| Daily | 8 (1.8%) | 9 (3.8%) | 0 (0%) | 17 (2.4%) |
| Total (N) | 445 | 238 | 19 | 702 |