What are Therapists’ Attitudes Towards the Use of AI in Therapy?: Implications for Mentoring









Prescott, J., & Hanley, T. (2023). Therapists’ attitudes towards the use of AI in therapeutic practice: considering the therapeutic alliance. Mental Health and Social Inclusion, 27(2), 177-185. https://doi.org/10.1108/MHSI-02-2023-0020

The use of artificial intelligence (AI) in mental health and other services has been a topic of growing interest and debate. Machine therapists–i.e., “conversational agents” that communicate with patients. Previous research has explored the potential benefits and challenges of incorporating AI into therapeutic practice, such as:

  • improving access to care
  • reducing stigma
  • freeing up mental health professionals and paraprofessionals’ time (Fiske et al., 2019; Stix, 2018).

But how do helpers feel about  using AI in therapeutic contexts?

In this study, researchers explored whether therapists thought it would affect the therapeutic alliance–the agreement between the therapist and client on the goals and tasks of therapy, as well as the emotional bond between them. This alliance has been consistently linked to successful therapy outcomes (Norcross & Wampold, 2018). As AI technologies continue to advance, it is crucial to understand whether therapists believe that the components of the therapeutic alliance can be achieved using AI in therapy.

The present study aimed to address two research questions:

(1) What are therapists’ attitudes towards the use of AI in therapy?

(2) Do therapists believe that the components of the therapeutic alliance (goal, task, and bond) can be achieved using AI in therapy?


An exploratory survey was developed to solicit the views of therapists and therapists in training about their attitudes towards the use of AI in therapy. Participants were recruited through convenience sampling, using a snowball strategy based on the authors’ contacts and associations within the profession. A total of 35 counselors completed the questionnaire, with 26 (74%) identifying as trainee therapists and 9 (26%) as qualified therapists. The majority (74%) practiced an integrated approach, while 14% practiced cognitive-behavioral therapy, and 11% practiced a psychodynamic approach. Participants completed an online questionnaire that included Likert-scale questions and an open-ended question. The questionnaire was based on the Working Alliance Inventory short version (Hatcher & Gillaspy, 2006), with items revised to consider whether the therapist thought that the component (goal, task, or bond) could be offered by a machine therapist. Additionally, the authors devised 12 questions to assess attitudes towards machine therapy in terms of access, general attitudes, risks, and training.


Attitudes Towards Artificial Intelligence

The majority of participants (56%) indicated that they would not consider accessing therapy by a machine therapist  while 35% selected “maybe,” and 9% selected “yes.” Participants generally disagreed that AI can help reduce social stigma around mental health, that AI is a promising therapeutic approach, and that a machine therapist can provide effective support. However, they tended to agree that professional standards are required in the use of machine therapy and that AI should only be used for training purposes, not for real therapy.

Therapeutic Alliance

Based on the mean scores for the subscales within the therapeutic alliance, the use of AI and machine therapy was viewed more positively for goals, followed by tasks, and lastly, the bond. The bond, being the more personal aspect of the therapeutic alliance, was the area of greatest concern for participants.

Qualitative responses were grouped into four themes: negative responses, possibilities of AI, mixed support, and therapeutic risks. Negative responses expressed concerns about the potential harm and ethical issues associated with machine therapy. Responses highlighting the possibilities of AI recognized its potential as a training tool and additional resource for therapy. Mixed support responses acknowledged both the potential benefits and limitations of AI in therapeutic settings. Responses related to therapeutic risks emphasized the importance of human connection, empathy, and the ability to recognize nuances in emotional expression, which participants felt might be lacking in machine therapy.


The findings revealed a clear reservation among therapists and trainee therapists about the use of AI technology within the therapeutic setting. These same reservations are likely to arise among mentors and other paraprofessional helps.However, many participants also recognized the potential and were open to the possibilities this technology could bring. The therapists’ largest concern was about the potential to create a strong bond with a machine therapist, which aligns with the importance placed on the therapeutic alliance in successful therapy outcomes.