Part One: 
Before reading the article, I honestly didn’t realize there were so many different forms of AI. I always pictured AI as the Google summary that pops up when I search something or people making AI pictures and videos. I never really stopped to think about how widely AI is actually used.

Once I started thinking about it, I realized I’ve actually encountered AI a lot more than I thought. Of course, the main one is Google Search because the AI summary is so convenient. I realized how much I’ve gotten used to having the AI summary because if it doesn’t appear, I’ll usually search it again a different way until it does. I’ve also noticed I use AI a lot on YouTube. Once I put two and two together, I realized it’s AI showing me comments that match the timestamp of the part of the video I’m watching. Alexa using AI was also a really surprising change when it first showed up. As we keep moving into the future, AI is expanding into phones, cars, drive-thrus, and it feels like it’s being added to almost everything.

Realizing how often I use AI also changed the way I think about it. Before reading this article, I knew AI wasn’t actually thinking like a person, but I don’t think I fully understood what that meant. I think I gave AI more credit than it deserved. Learning that it mainly recognizes patterns and predicts what should come next made it seem a little less “cool” to me. It also made me realize why AI can sound so confident even when it’s wrong. Just because it gives an answer that sounds believable doesn’t mean it actually understands what it’s saying.

Because of that, I think understanding how AI works makes me a little less comfortable relying on it in social work. I think it could be helpful for things like checking grammar or brainstorming ideas, but I don’t think it should ever replace a social worker’s judgment. Every client has a different story, emotions, culture, and experiences, and I don’t think AI can truly understand those things the way a person can.

If someone asked me what an AI hallucination is, I would probably explain it by saying it’s when AI gives an answer that sounds completely right and says it with confidence, but the information is actually wrong or made up. AI isn’t trying to lie, it just predicts what it thinks the answer should be based on patterns it has learned. That’s why I think it’s important to double-check important information instead of assuming AI is always correct.

Part Two:
After reading Moore et al., I don’t think the answer is to completely get rid of AI in social work because I don’t think that’s realistic anymore. AI is already becoming part of healthcare and a lot of other jobs, so I think the better question is how we can use it responsibly. Reading about the stigma and some of the unsafe responses AI gave made me realize that AI can say the wrong thing or misunderstand someone’s situation if there isn’t enough human oversight.

I think the biggest thing I would want is for a social worker or another trained professional to always review what AI is doing before it affects a client. I think AI could be helpful for things like checking grammar, brainstorming ideas, creating educational handouts, or answering basic questions after hours. Those are things that could save social workers time. But if someone is talking about their mental health, trauma, or anything that could affect their safety or well-being, I don’t think AI should be handling that by itself. Every client has their own experiences, culture, and emotions, and I don’t think AI can truly understand those things.

Working in behavioral health has also made me think about this differently. Sometimes I think that if AI were in charge of assessments, there would probably be a lot more people on involuntary holds. That’s obviously an exaggeration, but it shows why I don’t think AI should be making those decisions on its own. I’ve seen patients come into the emergency department with similar complaints, but once you actually start talking to them, you realize their situations are completely different. Two people might both come in because they’re feeling depressed, but the reasons behind it, what they need, and how they’re feeling can be completely different. There are so many things you learn by actually talking to someone that a computer can’t pick up. That’s why I think there always needs to be a person reviewing things because it’s easy to miss something important if you’re only looking at information on a screen.

Overall, I think AI should be used as a tool to help social workers, not replace them. If it can save time on simple tasks, that’s great because it gives social workers more time to focus on their clients. But I don’t think AI will ever replace empathy, professional judgment, or the relationship that develops between a social worker and a client. After reading Moore et al., I actually think those human parts of the job are even more important because AI can sound very convincing, even when it’s wrong.