ENG3702 Assignment 2
(COMPLETE ANSWERS)
2025 - DUE 21 July 2025
For assistance contact
Email:
, The Illusory Impartiality: Deconstructing "Neutrality" in AI-Powered Therapy for Non-
Professionals
In an era defined by unprecedented technological accessibility, the proliferation of Artificial
Intelligence (AI) chatbots such as ChatGPT, GROK, Copilot, and Gemini has extended their
reach into domains traditionally reserved for human expertise. Among these burgeoning
applications, the casual and indiscriminate use of Gen-AI as a therapeutic tool by non-
professionals presents a profound and often unrecognized challenge. This essay argues that the
perceived "neutrality" of these AI applications, when deployed in the sensitive realm of mental
health support, is an dangerous illusion. This illusion stems from the inherent biases embedded
within the very fabric of language, text, genre, and discourse, profoundly impacting the
therapeutic relationship and potential outcomes. By critically examining how AI constructs
discourse, caters to its audience, and inadvertently perpetuates non-neutral perspectives, this
essay will demonstrate the significant implications of relying on AI for therapeutic intervention
without professional oversight.
The discourse generated by AI is far from a neutral transmission; it is a meticulously constructed
artifact shaped by the interplay of language, text, and genre. As discussed in Online Session 8,
language is not merely a tool for communication but a system imbued with cultural, social, and
historical contexts. AI models, trained on colossal datasets of human-generated text, inevitably
internalize and reflect the linguistic patterns, societal norms, and even the biases prevalent within
that data. The specific "text" the AI produces – the words, phrases, and sentence structures it
selects in response to a user's prompt – is therefore a product of this biased training. For instance,
an AI might inadvertently adopt a common, yet potentially stigmatizing, discourse around certain
mental health conditions, reflecting prevalent online discussions rather than clinically nuanced
perspectives. This is not a neutral act but a reproduction of existing linguistic frameworks.
Furthermore, AI, even when attempting to simulate therapeutic conversation, implicitly adopts
elements of various "genres." It might blend informational exposition, supportive conversational
cues, and problem-solving frameworks. These genre conventions, often unconsciously, dictate
the AI's responses and shape the user's perception of the interaction. The convergence of this
language, text, and genre creates a specific "discourse" – a way of understanding and talking
about mental health that is fundamentally influenced by the AI's programming and the nature of
its training data. Consequently, the very mechanisms by which AI generates discourse preclude
true neutrality; the selection of training data, the underlying algorithms, and the inherent
structures of language itself introduce a non-neutral bias, even if unintentional.
Beyond its internal construction of discourse, AI applications are meticulously designed to cater
to their "audience" – the user – through specific communicative strategies, further complicating
claims of neutrality. Firstly, AI attempts to establish an "interpersonal relation" with the user,
often employing empathetic language, active listening cues, and supportive phrasing. This
engineered "relation," while designed for engagement, can be perceived as genuine by non-
professional users seeking therapeutic support, fostering a false sense of trust and understanding.
However, this manufactured connection is not neutral; it is a calculated response aimed at
eliciting a certain user interaction, potentially obscuring underlying algorithmic biases or
(COMPLETE ANSWERS)
2025 - DUE 21 July 2025
For assistance contact
Email:
, The Illusory Impartiality: Deconstructing "Neutrality" in AI-Powered Therapy for Non-
Professionals
In an era defined by unprecedented technological accessibility, the proliferation of Artificial
Intelligence (AI) chatbots such as ChatGPT, GROK, Copilot, and Gemini has extended their
reach into domains traditionally reserved for human expertise. Among these burgeoning
applications, the casual and indiscriminate use of Gen-AI as a therapeutic tool by non-
professionals presents a profound and often unrecognized challenge. This essay argues that the
perceived "neutrality" of these AI applications, when deployed in the sensitive realm of mental
health support, is an dangerous illusion. This illusion stems from the inherent biases embedded
within the very fabric of language, text, genre, and discourse, profoundly impacting the
therapeutic relationship and potential outcomes. By critically examining how AI constructs
discourse, caters to its audience, and inadvertently perpetuates non-neutral perspectives, this
essay will demonstrate the significant implications of relying on AI for therapeutic intervention
without professional oversight.
The discourse generated by AI is far from a neutral transmission; it is a meticulously constructed
artifact shaped by the interplay of language, text, and genre. As discussed in Online Session 8,
language is not merely a tool for communication but a system imbued with cultural, social, and
historical contexts. AI models, trained on colossal datasets of human-generated text, inevitably
internalize and reflect the linguistic patterns, societal norms, and even the biases prevalent within
that data. The specific "text" the AI produces – the words, phrases, and sentence structures it
selects in response to a user's prompt – is therefore a product of this biased training. For instance,
an AI might inadvertently adopt a common, yet potentially stigmatizing, discourse around certain
mental health conditions, reflecting prevalent online discussions rather than clinically nuanced
perspectives. This is not a neutral act but a reproduction of existing linguistic frameworks.
Furthermore, AI, even when attempting to simulate therapeutic conversation, implicitly adopts
elements of various "genres." It might blend informational exposition, supportive conversational
cues, and problem-solving frameworks. These genre conventions, often unconsciously, dictate
the AI's responses and shape the user's perception of the interaction. The convergence of this
language, text, and genre creates a specific "discourse" – a way of understanding and talking
about mental health that is fundamentally influenced by the AI's programming and the nature of
its training data. Consequently, the very mechanisms by which AI generates discourse preclude
true neutrality; the selection of training data, the underlying algorithms, and the inherent
structures of language itself introduce a non-neutral bias, even if unintentional.
Beyond its internal construction of discourse, AI applications are meticulously designed to cater
to their "audience" – the user – through specific communicative strategies, further complicating
claims of neutrality. Firstly, AI attempts to establish an "interpersonal relation" with the user,
often employing empathetic language, active listening cues, and supportive phrasing. This
engineered "relation," while designed for engagement, can be perceived as genuine by non-
professional users seeking therapeutic support, fostering a false sense of trust and understanding.
However, this manufactured connection is not neutral; it is a calculated response aimed at
eliciting a certain user interaction, potentially obscuring underlying algorithmic biases or
