Part 1 – Kalota Priomer
One specific concept from this article that changed what I previously believed about AI is the growing involvement of AI usage within my own life experiences. To explain further, at the start of this course, I had a firm stance that I do not engage in AI, however Priomer’s article mentioned how this technology can be found in multiple different areas. The place that I did not consider the most is seen in LLM that can be found in emails and text messages which are two forms of communication that I use very frequently. Therefore, this has challenged my personal belief that I do not engage in AI usage.
Personal AI Encounters
I have encountered generative AI within my own personal life unknowingly on several occasions. My previous interpretation of generative AI consisted of applications such as ChatGPT. Understanding the other more discreet forms of AI, I can admit that AI is very involved in my technology usage. As mentioned, I have taken advantage of text and email word suggestions as well as the predicted text options when writing several assignments for the social work program. I have also unknowingly used AI when making simple searches. It is common to see on multiple platforms the summarization of searches and the content that is displayed by AI tools. An example of this would be google searches as it now uses AI to communicate search results.
Comfortability with AI in Social Work Practice
Understanding how AI works makes me more comfortable with AI showing up in social work practice as I have a better grasp on where it is present and the different ways it is used. As mentioned within the Priomer (2026) article, AI can aid creativity and assist in refining ideas. Upon review I can comprehend that AI can be beneficial in being innovative by fostering new ideas. I believe being innovative is necessary to be an effective social worker and leader within my community and AI can generate media and quick solutions. Acknowledging that this system could be overused or misused in certain instances when original work or when privacy of clients could be compromised, I believe a large responsibility falls upon the social worker to use it appropriately in practice. I also found it worth noting disadvantages when determining whether AI makes me more or less comfortable in social work practice. Some disadvantages that are most worrisome in regards to AI usage is overutilization in health and mental health fields, inaccurate data output, possible privacy violations, and environmental sustainability issues. While these are big issues that create concern, I feel that in a society where technology is continuing to grow, it is necessary to embrace AI and create policies and solutions for these problems.
Why would AI hallucinate?
An AI tool “hallucinates” because it is still in the early stages of this technology and relies on imputed data and information it has received. Currently it is only available as Artificial Narrow Intelligence which means it can only perform limited tasks and does not have the capacity to learn on its own or gain its own beliefs. Essentially, if the data being input into AI is inaccurate or contains bias, it will provide that same inaccurate information as it can not yet learn outside data.
Part 2 – Moore et al.
Safeguards
To state clearly, one safeguard I’d suggest before being okay with the tool being used with real clients is regulation. As stated within the Moore et al. (2025) article, there is a rise in “wellness apps” that many are using for mental health services. It is also noted that these apps are not regulated by state or federal government and should not replace the role of therapist but instead aid therapists.
Where to draw the line?
To draw a line between an appropriate task for AI and a task that requires a human, I would suggest limiting AI to administrative duties and to complete tasks that do not work directly with clients. After working in behavioral health, I can understand how completing progress notes, organizing information, screening potential clients, and creating templates can be time consuming. I feel that these are examples of noninvasive and appropriate tasks for AI. I believe that communicating with clients, creating treatment plans, and diagnosing clients should be left for therapists. I believe Moore et al.’s findings helped me draw this line as he describes the effects of individuals receiving the same quality of care from AI as going to low quality professionals. I also referred to the study that describes how AI systems have even shown stigma towards certain mental health conditions such as alcoholism. This has led me to believe that AI should not work directly with clients as pushing against stigmas plays a major role in the difference between an ineffective and effective therapist. Another finding that has helped draw the line was the study that showed the limitations AI has for responding appropriately with assisting individuals with suicidal ideations. This study showed that the AI service lacked empathy and was very dismissive. Overall, I feel AI lacks empathy and can not effectively imitate empathy which is essential when working with groups and individuals battling mental health concerns.
What happens when this oversight is missing?
Within my workplace environment, I have witnessed how the lack of accurate information and the lack of empathy for individual circumstances can have a negative effect on work quality, burnout, and mental health. These oversights have caused individuals to feel uninformed about expectations while also feeling unsupported by leadership. While I have not witnessed this as a result of AI, I imagine AI could cause the same effects which may include Increased stress and frustration, low morale, decreased organizational commitment, more mistakes and reduced work quality, and increased turnover due to poor mental health advocacy in the workplace. Simultaneously, I can see the benefits of using ai to manage workflow and alleviate high caseloads through administrative tasks. Not having systems in place for these concerns can also lead to the same low morale, stress, and frustrations that many professionals experience within my workplace. Understanding that two ideas can exist concurrently, I feel this is what helped establish the oversight for where to draw the line on AI usage and we should embrace this technology with limits.
Tia,
Your acknowledging that you understand AI better is an important first step. And when we better understand how something works, we are more able to see how it might be useful to us with the correct safeguards. In my document that I added to this week’s module, I have differentiated between AI and Gen AI. In your first examples, you are probably using AI, which is predictive, while ChatGPT is Gen AI, which creates new content. If you look at the document on What type of AI, it will give you more explanation. I think we are all a bit surprised when we truly look at the ways AI has infiltrated our everyday technology use.
You have accurately defined the problem – finding the line where AI is useful, but not allow it to be responsisble for things that would be harmful. The problem is decided where that is and then creating the safeguards that could prevent problems.
Good discussion.
Dr P