For this reflection, I listened to Timnit Gebru’s podcast “Don’t Fall for the AI Hype” and watched the Model Employee video. The common thing between them is that they both made AI look less simple than I usually think of it. It is not just a tool that helps us write, search, or generate things. There is data behind it, companies behind it, and also people who can be harmed by it. Since I already use AI sometimes for university work and design, the topic felt close to me and not just something far away.

One thing that shocked me in the podcast was the example of Joy Buolamwini and facial recognition. Timnit Gebru talked about how Joy showed that some face detection systems could not recognize her face unless she wore a white mask. This was shocking because it made AI bias feel very clear. It is not just a word we say in class. It is a machine literally failing to see someone properly because of their skin tone. If this happens in a phone camera it is already bad, but if it happens in airports, security systems, policing, or work places, then it becomes dangerous. It showed me that the people who build these systems and the data they use really matter.

Another thing that surprised me was how companies talk about AI like it has its own brain. I did not really notice this before, but Timnit Gebru made me think about it. When people say “the AI decided” or “the AI made a mistake,” it sounds like no human is responsible. But AI did not appear by itself. People chose the data, trained the model, released it, and made profit from it. So if it gives a racist result or harms someone, the blame should not just go to “the machine.” The company and the people behind it should also be questioned.

The part about Timnit herself being pushed out of Google was also strange to me. She was working on AI ethics, but then she got in trouble for speaking about the risks of AI models. This felt ironic. It made me think that maybe big companies like ethics when it makes them look good, but not when it actually challenges the product or the profit. This part made AI feel more connected to power, not only technology.

The Model Employee video shocked me in another way. I already knew that apps collect data, but seeing it in the video made it more uncomfortable. It showed how data can be used to watch people, especially at work. It is not just about collecting your email or your name. It can become about tracking your productivity, movement, habits, and judging you through this information. I used to think sometimes “I have nothing to hide,” but this video changed that a bit. Privacy is not only about hiding something wrong. It is about still having control over your own information.

This connects to what we did in class about terms and conditions, privacy policies, and digital footprints. We talked about how most people press accept without reading, and honestly I do that most of the time. I do not read the whole thing because it is too long and I just want to use the app. But after Model Employee, the accept button feels different. It can mean I allowed the app to collect things from my phone, my location, my photos, or my activity without me really thinking about it.

It also reminded me of what we said in class about language and AI answers. The answer does not only depend on the question, but also on the language I use. For example, if I ask AI about the Arab world in English, sometimes the answer feels like it is written from a Western side, or like someone outside the region is explaining us. But if I ask the same thing in Arabic, the answer can feel closer to the Arab context. Maybe it understands the tone more, or maybe it gives more sympathy to the Arab side. So bias is not always obvious. It can also be in the language itself, and in whose point of view is treated as the normal one.

Something I will do differently is how I use AI in architecture and design. I can still use it for simple things, like organizing my ideas or helping me explain a sentence better. But I do not want to depend on it in parts that need real site research. For example, if I am designing for a neighborhood or a group of users, I cannot just ask AI what people need and take it as my answer. This is too easy, and it skips the real process. People’s needs depend on the site, their daily life, their habits, and what they actually say when I go and ask them. AI can guess, but it cannot replace going to the site, observing, and talking to people myself.

I will also be more careful with personal photos and private information. Before, AI photo trends looked normal to me because everyone was trying them. But now I feel they are not that safe. I heard many times that people can use someone’s photo to generate fake images, and some of these tools are free and easy to use. This is scary because maybe someone can use my face for something I never agreed to. It is not only photos too. AI can maybe copy someone’s writing style or create a fake chat that sounds like them, then use it to trick their friends or people they know. So now uploading personal things online feels more serious to me.

One thing I think the course can add is a practical class about apps we already use. For example, we can look at TikTok, Instagram, Google, or AI image tools and see what they actually ask for. Like camera, microphone, location, photos, contacts, or saved activity. I think seeing it directly would be better than only saying “apps collect data.” Because honestly, we all know they collect data, but we do not always know what exactly we accepted.

I also think the course should talk more about the people behind AI. Before, I did not really think about the workers who label data or review harmful content so the AI looks clean for us. I used to imagine AI as just a system working alone. But there are people behind it doing difficult work, and sometimes this work can affect their mental health. This part is usually hidden, but it matters because it shows that AI is not only technology. There is labor behind it too.

After the podcast and the video, I still think AI is useful, but I do not see it the same way anymore. I will still use it, but with limits. I want to be more careful with what I trust, what I upload, and when I should do the work myself instead of letting AI make it easier for me. Especially in architecture, I do not want AI to replace the part where I go to the site and listen to real people.

I used Grammarly to correct my grammar and my writing