When my 2012 article Narrative techniques in reflective practice got a plug in this post from NICEC, I started wondering how it would have been different if I had written it today in the age of AI and auto-transcribed online meetings.
Whilst the prevailing paradigm in AI usage for many is to see it as some magical black box generating outputs in seconds to save you the effort of thinking, other people have been exploring how AI could be used in more sophisticated ways to facilitate deeper thinking.
So, I uploaded my article to Gemini and got it to brainstorm some ideas for how AI could be incorporated into some of the narrative reflective techniques mentioned. I then then used that as a prompt to come up with my own (better) list of ideas.
Bear in mind that these are just speculative ideas and that AI is likely to be rubbish at some of these things because it's not good at contextualised nuance. You will need to evaluate the validity and applicability of any responses. However, the real aim is to try to get the AI to respond in ways that make you think about your own practice differently. So, even if it's wrong, it still might be useful.
WARNING:
Think at least three times before uploading any personally identifiable or sensitive information to any AI product — then think again.
NEVER use client information or session transcripts in this way (even if anonymised and redacted) without the client's informed consent.
Thematic analysis
You could upload transcripts or summaries of one or more interactions and ask your AI tool to analyse the client narrative and identify (with illustrative excerpts):
- any recurring or dominant themes
- any potential themes that could have been more fully explored
Self-confrontation
- Get the AI to analyse the transcript and identify distinct joint sub-projects that the you and the client appeared to be working on together. Then get it to estimate the apparent degree of alignment between your goals and approaches.
- Describe your goals for the session at various points. Then get the AI to look for your actions that were consistent or inconsistent with those stated goals. Ask it to suggest other possible interpretations for your actions.
- Share your perceptions of the clients goals at various points. Get the AI to look for evidence in the client's actions that would support or challenge your perceptions.
- Get the AI to play the role of the client providing a commentary on the underlying reasons for their own contributions and on their reactions to your contributions.
Archetypes
- Get the AI to identify evidence for or against the presence of certain archetypal plots in the client's narrative, and to discuss the possible implications.
- Get the AI to retell the client narrative to fit with a chosen archetype and then review the similarities and differences between the original and the retelling.
- See here for some interesting archetype frameworks
Relating to theory
- Feed in a summary of a career development theory and ask the AI to identify possible indicators of how relevant the themes in that theory would be to the client's narrative, and where it might be misleading if it were applied too rigidly.
- Get the AI to generate an alternative version of how the discussion would have gone if a particular theory had been applied consistently to interpret the client's input. Ask for a best case and a worst case version.
Structural analysis
You could ask the AI tool to provide a bullet point summary or flow chart of the structure of the client narrative (or the narrative of the discussion). You could ask it to identify:
- the main plot points and transitions
- key characters and their roles in the narrative
- the use of narrative devices such as metaphors, imagery or flashbacks
Actant analysis
- Get AI to apply Greimas' actant model to the client's narrative identifying individuals or other entities that may have played those actant roles, as well as cataloguing any missing actants. Remember, an actant doesn't have to be an actual person — it can be an idea or an aspect of someone's personality.
- Tell the AI to identify the client as the Subject and then examine which of the other actant roles you played throughout the interaction. Get it to suggest other roles you could have played at various points.
- Get the AI to retell the narrative of the discussion with you as the Subject. Ask it to identify the other actants in this narrative.
Action identification and perspective
- Get AI to chart the level of construal (from low-level concrete description to high-level abstract interpretation) evident in the language of both client and practitioner throughout the interaction and explore the degree to which these were matched or mis-matched at various points.
- Ask AI to identify any potential misunderstandings resulting from mismatched construal and to suggest alternative responses at a similar level.
- Get AI to suggest alternative practitioner responses and questions at each point that could have shifted the discussion to higher or lower construal levels.
- Ask AI to produce three critical summaries of the interaction from different perspectives — the practitioner's, the client's and a neutral third person observer — the summaries could include questions each person might want to ask about the discussion at various points.
Interactional analysis
You could ask AI to analyse the interaction between practitioner and client, identifying patterns in the communication. You could get AI to look for:
- the extent to which your approach compares to recommended quality standards or ethical principles
- effective or ineffective use of particular counselling techniques such as paraphrasing, restating, self-disclosure, etc.
- habits in the use of different types of questions (closed, open, leading, multiple, etc.) and suggest alternatives
- patterns in the client's communication and to speculate on what that might mean about their underlying needs and ways of processing information and approaching challenges.
Valuation theory
- Get AI to classify client and practitioner contributions according to the extent to which they show evidence of motivations for 'self-enhancement' and/or 'contact/union with others'
- Get AI to characterise the possible emotional content of each contribution
Dialogical self
- Get AI to suggest examples from the client's contributions when they may be speaking from different I-positions (e.g. Super's life roles, or Wheel of life) and identify possible evidence of harmony and conflict between these positions.
- Identify possible I-positions you may have taken (e.g. 'professional helper', 'thinking partner', 'motivation builder') and get AI to rewrite your contributions so that they are all consistent with this I-position.
Performative analysis
If your AI do sentiment analysis on audio or video, you could use it for that. However, most can't yet (and even if they could, remember the confidentiality warning). However, you could:
- Get it to analyse a transcript and how various non-verbal cues might shift the underlying meaning of the client's statements, then compare these to the client's actual non-verbal signals.
- Feed AI a client's narrative and then ask it to role-play the client for you. You could mindfully record your feelings at each point. You could go back and re-run the simulation from any point to see the impact of taking different approaches. You can even ask the AI to step out of character and ask you questions to prompt mindful reflection or provide suggestions for how to move forward at certain points.
