Practice Notes: AI: Friend or Foe?

Practice Notes: AI: Friend or Foe?

Practice Notes is an educational tool designed to help Ontario social workers, social service workers, employers and members of the public gain a better understanding of recurring issues dealt with by the College’s Professional Practice Department and Complaints Committee that may affect everyday practice. The notes offer general guidance only and College registrants [1] with specific practice inquiries should consult the College (or their lawyer for any legal issues), since the relevant standards and appropriate course of action will vary depending on the situation.

Registrants are contacting the Professional Practice Department with questions about the use of Artificial Intelligence (AI): How can it be used in practice? What are the risks? And, most importantly, what do the Code of Ethics and Standards of Practice (the Standards) require?

As the use of AI becomes more common and embraced in our professional lives, registrants should recognize that decisions and actions once grounded in research, analysis, and dialogue may now be influenced by information generated by technology. The Professional Practice Department has heard that some registrants have proceeded cautiously in using AI while others have been more willing to implement various aspects of AI. Registrants want to know how the Standards apply and are seeking information from the College in order to ensure that they are practising ethically and competently.

As a baseline, the College’s expectation, in all aspects of practice, is that registrants conduct themselves in accordance with the Standards regardless of the technologies that they use in their practice. This means that registrants have a responsibility to both understand their professional obligations and be competent in whatever technologies they are using to meet those obligations.    

The College does not recommend particular AI tools, nor does it provide legal or technology advice. It is the registrant’s sole responsibility to identify the potential risks associated with any technology they are using and address those risks to meet their professional obligations.  What the College can do is help registrants understand how the Standards can arise or apply when AI or any technology is being considered and used. Registrants must also consider their own strategies and protocols for the use of AI as well as other requirements that may apply in their workplace or practice setting and seek consultation with specialists in the technology and legal sectors when needed.

This resource uses practical scenarios to explore themes about the importance of remaining accountable and professional in the presence of AI. Throughout, the focus is on the fact that using AI does not change a registrant’s professional responsibilities. Specifically, among other expectations, obligations to provide confidential, accurate and relevant services to clients are discussed within the context of scenarios registrants may find themselves challenged by.

Scenario One – Use of AI and the Client Record

A registrant contacts the Professional Practice Department after their workplace introduces an AI tool to scribe client sessions and integrate information into the client record. They asked what the Standards require in this situation.

The Professional Practice staff noted that the use of an AI scribe to capture and synthesize information from a counselling session with a client may affect one of the most significant of rights of clients, their privacy rights.  When AI is introduced to summarize specific client information, AI platforms may take information that is inputted and use it for training its system or other unknown uses. Safeguarding of personal data and respect for an individual’s privacy must be prioritized by registrants. Clients should be made aware of how registrants may use AI and registrants will want to share with clients what workplace policies exist to protect client information and whether they can decline AI use in their care.

The Professional Practice staff shared information found in both Principle IV: The Social Work and Social Service Work Record and Principle V: Confidentiality [2] in the Standards which describe a registrant’s responsibilities regarding records, confidentiality obligations, and that consent from clients must be obtained when collecting, using and disclosing their personal information, with limited exceptions. Principle III: Responsibility to Client [3]  was also reviewed and the registrant was reminded that they must be transparent in describing their services and do so in a manner that clients understand what benefits or risks may result in using their services.

It was also suggested that the registrant review the College’s Privacy Toolkit to further their understating of privacy legislation, and to consider a resource jointly developed by the Information and Privacy Commissioner of Ontario (IPC) and the Ontario Human Rights Commission (OHRC) on AI use [4], as well as recommendations from the Privacy Commissioner of Canada regarding inputting information into AI platforms.[5]

In this scenario, the Professional Practice staff also discussed that AI generated information may include “hallucinations” and non-existent data.  Record integrity may be compromised as AI may add inaccurate information into session notes when it recognizes a pattern in a sentence it has noticed with similar data elsewhere in its training models. The registrant was referred to Principle IV: The Social Work and Social Service Work Record: 4.1.4 [6] where it explains that registrants must ensure they are not documenting information they know may be false or misleading.

After consulting with the Professional Practice Department, the registrant decided to check with their workplace on their policies on the use of an AI scribe. They will engage with their workplace to support building processes that value client privacy rights, transparency and record integrity.

Scenario Two – The Use of AI and System Bias

A registrant works for a large city supporting social policies that aim to benefit local communities. They have been asked to develop an affordable housing plan for vulnerable individuals. They are told to use AI to acquire information about various neighbourhoods, such as what services and resources exist, as well as general demographics about individuals living in those neighbourhoods. The registrant is concerned this process will not truly capture the needs of the individuals seeking affordable housing. To make matters more worrisome for the registrant, the future needs of residents were a subject of significant concern at a recent public consultation. The registrant inquires about their work in the context of the Standards.

Professional Practice staff raised the concern that inherently, AI is broadly understood to be adaptive and will learn and evolve over time as users provide new inputs to the models. This means that AI results can vary over time, even with the same inputs, as well as hallucinate information.

Registrants have an obligation to ensure that any professional recommendations or opinions they provide are supported by evidence and a credible body of professional social work or social service work knowledge (Principle II: Competence and Integrity [7]). For registrants, it can be challenging to know for certain if sources of information from AI are valid, or why certain information is provided by an AI platform. Registrant should approach AI generated information with caution and it not a substitute for registrants exercising good, professional judgment. With this in mind, registrants should instead consider the intended goals and outcomes for their clients and focus on appropriate client-focused interventions. Professional Practice staff shared the following relevant Standards of Practice:

Principle I: Relationship with Clients:

1.2 Goals for relationships between College registrants and clients shall include enhancing clients’ functioning and strengthening their capacity to adapt and make changes. [8]

1.3 College registrants shall observe, clarify and inquire about information presented to them by clients.[9]

Principle II: Competence and Integrity:[10]

2.1.6 and 2.1.7 emphasizes that registrants should be aware of and understand the population they are serving and competently equip themselves with accurate information and evidence when making decisions.

The Professional Practice staff and the registrant discussed that reliance on AI can mean missing crucial elements of their work. For example, important local knowledge may be lacking, equity-denied groups may be omitted from the planning, or desired community hubs or services may not be considered.  It is widely acknowledged that AI risks proportionately affect “Indigenous, racialized or otherwise vulnerable communities and individuals”. [11] The more there is dependence on AI, the further the lack of relationship with clients, colleagues and systems which can impact service delivery.

The registrant recognized that depending solely on AI-generated information can overlook important context, nuance and the lived experiences of individuals and communities. Moving forward, they will search data from reliable and relevant sources and engage with the community and residents as they work on the housing plan. They realized that without these steps, issues like fairness, equality and equity might be lost. The registrant understood that it is important not to lose touch with being the “human in the loop” – where rapport and trust is built and where credibility, compassion and concern lie.

Scenario Three – AI and Professional Guidance:

A registrant asked the Professional Practice Department about accessing AI tools in lieu of consultation/supervision/legal consultation because it is cheaper and more accessible. They wondered whether they could enter non-identifying information about a case and receive guidance on legislation, modalities and treatment plans or get feedback on their clinical judgment.

The registrant and Professional Practice staff reflected on how important it is not to assume that AI–generated information is accurate, applicable and current. Making these assumptions could reduce effective and individualized approaches for clients and may even lead to inappropriate or unprofessional services. Ultimately, registrants are required to seek supervision, where appropriate, in order to protect their clients and that supervision is required to be meaningful.  When information is inputted into an AI platform it may not account for such things as client body language, non-verbal patterns of behaviour, relational dynamics or other issues that might be relevant in the circumstances. Misaligning services with the client’s needs and best interests by AI use can occur at all stages of service, from initial contact to final outcomes.

Further, working in isolation in this way creates no space for new, inventive ways to support clients or for building different models of care or service that may enhance the registrant’s practice. This is due to the fact that AI is known to perpetuate andsupply information that validates and encourages a pattern, a line of thinking or existing practices rather than debating or challenging them. 

Principle II: Competence and Integrity: 2.1.5 and 2.1.8 [12] in the Standards indicate that registrants must be committed to ongoing professional development, evaluate their practice and seek supervision and consultation when appropriate.

The registrant concluded that they may not be able to identify the source of information to support their clinical work if obtaining consultation or supervision through AI, or whether the information accurately reflects their clients needs.  They also appreciated that AI does not have the ability to critique their conduct in a way that improves their competence nor does it generate options solely for the best interests of each of their clients, or the profession in general. The registrant agreed that they should continue to consult with supervisors, their mentors and communities of practice to reduce risks to clients.

Conclusion

Using AI does not change a registrant’s responsibility to practise ethically and competently and in accordance with the Standards.  How and when to use AI, or not, depends on many factors. Reflecting on the Standards, reviewing College resources such as The Use of Artificial Intelligence in Practice, and seeking guidance provided by privacy specialists will support good practice. 

Of special note, the College does not regulate organizations that employ social workers and social service workers. Consequently, the policies of organizations may not align with the College’s expectations – so registrants must be vigilant and not complicit in a breach of the Standards. Registrants are accountable for their conduct and, if their organization’s AI use conflicts with ethical practice, registrants need to advocate for workplace policies that align with the Standards as noted in Principle II: Competence and Integrity: 2.2.18 [13].  

Consider the following before using AI:

  • Do I know what AI does with the information I give it, and where that information is stored?
  • Have I obtained client’s informed consent if client information is put into an AI tool, and can the client decline without any effect on their service?
  • Am I reviewing and correcting the output received from AI before it becomes part of the client record or a report?
  • Can I defend my professional judgment behind my decisions, independent of information obtained through AI?
  • Have I verified any citations, legislations etc. that may have quoted before applying it in my work?
  • If I am being obligated to use AI by an agency or employer, and I have concerns have I documented and raised these concerns?

As seen in the scenarios, the use of AI and reliance on it have become common in many workplaces. Registrants may be asked to implement AI tools to improve efficiencies, and they may also be considering their own ways to benefit from using AI without assessing the risks or deficits. For these reasons, registrants will want to turn their minds to appreciate decision-making rooted in responsible practices. Registrants make decisions every day that impact clients and they should be able to be explain their decisions in the context of transparency, sound research, and meaningful practices.


[1] Disclaimer: the term “member” and “registrant” are used interchangeably and synonymously as equivalent to the term “member” as used in the Social Work and Social Service Work Act, 1998, and the Regulations.

[2] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle V: Confidentiality, Interpretation 5.1.2

[3] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle III: Responsibility to Clients, Interpretation 3.3

[4] https://www.ipc.on.ca/en/resources/principles-responsible-use-artificial-intelligence

[5] https://www.priv.gc.ca/en/privacy-topics/technology/artificial-intelligence/ai-chatbots_ind/

[6] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle IV: The Social Work and Social Service Work Record, interpretation 4.1.4.

[7] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle II: Competence and Integrity

[8] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle I: Relationship with Clients, Interpretation 1.2

[9] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle I: Relationship with Clients, Interpretation 1.3

[10] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle II: Competence and Integrity, Interpretations 2.1.6 and 2.1.7

[11] https://www.lco-cdo.org/wp-content/uploads/LCO-Federal-AI-Strategy-Submission-Final-Oct-31-2025.pdf

[12] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle II: Competence and Integrity, Interpretation 2.1.5 and 2.1.8

[13] (OCSWSSW), Code of Ethics and Standards of Practice, Third Edition, 2023, Principle II: Competence and Integrity, Interpretation 2.2.8