Part 1 – Kalota’s Primer

One thing that changed how I think about AI was learning that it does not actually “think” or understand information the way people do. Before reading Kalota, I thought AI searched for the correct answer like Google. Now I understand that it predicts the next word based on patterns it learned from a huge amount of data. That helped me understand why AI can sound so confident even when it is wrong.

I had a professor one semester that wanted us to use generative AI for school assignments, brainstorming ideas, and organizing our thoughts, it was actually required for our assignments. I have also seen people use AI to answer everyday questions and write emails. At my placement working with foster and adopted youth, I have thought about how AI could help create activities or educational materials, but I would still want to review everything before using it.

Understanding how AI works makes me a little more comfortable with it because I know what it can and cannot do. At the same time, I am more careful because I know it can make mistakes. If a client or coworker asked me why AI sometimes “hallucinates,” I would explain that AI is predicting what words are most likely to come next instead of checking if the information is true. Because of that, it can give an answer that sounds believable but is actually incorrect. That is why people should always verify important information instead of assuming AI is always right.

Part 2 – Managing, Not Just Rejecting, AI in Practice

After reading Moore et al., I do not think AI should replace social workers, especially in mental health. The article showed that AI can give biased, stigmatizing, or unsafe responses, which could seriously affect clients. However, I also do not think we should completely reject AI because agencies are already starting to use it. Instead, I think it should only be used with strong safeguards.

One safeguard I would require is that every AI response affecting a client be reviewed by a trained social worker before any important decision is made. AI could be helpful for lower-risk tasks like providing general psychoeducation, helping clients find resources, or answering basic questions after business hours. However, decisions involving suicide risk, crisis intervention, diagnoses, safety planning, or treatment recommendations should always be handled by a human professional.

My placement has shown me how important human oversight is. Every youth has different experiences and needs, and one response does not fit everyone. A computer cannot understand a person’s emotions, trauma, or situation the way a trained social worker can. Moore et al. reinforced my belief that AI should be used as a tool to support social workers, not replace their professional judgment, empathy, or decision-making. As AI continues to grow, I think social workers need to stay involved so it is used responsibly and ethically.

Reference:

Kalota, F. (2024). A primer on generative artificial intelligence. Education Sciences, 14(2), 172. https://www.mdpi.com/2227-7102/14/2/172

Moore, J., Grabb, D., Agnew, W., Klyman, K., Chancellor, S., Ong, D.C., & Haber, N. (2025). Expressing stigma and inappropriate responses prevents LLMS from safely replacing mental health providers. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT’25) (pp.599-627). Association for Computing Machinery. https://dl.acm.org/doi/epdf/10.1145/3715275.3732039