I answer medical questions for a voice AI. I’m scared about what it wants to do with the info.

I work for a large company, which I’ll call ClearVoice. The office itself is about half a kilometer square and I’ve never seen anyone from management or HR in person. <br><br>I feel like I can say this because I don’t know what department I’m working in, or even what the building looks like overall. I don’t have any way of knowing which part of it I’m in. I was brought here by a man wearing a guard uniform when I graduated University. My mom died when I was young, and my dad’s in a facility with frontotemporal dementia. My wife and daughter died in a car accident last year.<br><br>I accepted the job offer without asking any questions because, given my situation, I wasn’t in the headspace to be picky about what was happening. Everything seemed really professional. A man in a uniform took me to my station, then I didn’t leave for the rest of the 8 hours, and I was only let out to go to my apartment at the end of the day.<br><br>Once I was there, they told me what my job was: I would be given audio samples and I would have to answer questions. The samples would be of patients and doctors talking in a clinical setting. It was all from the patient’s perspective, so I couldn’t hear the actual doctor, but I could hear the audio of them talking to the patient. I’d be given a list of questions about what they were saying, things like “What is the purpose of the appointment?” “What diagnosis did they give?” “What are the doctor’s instructions for the patient? ?”<br><br>I thought I was training an AI or something because I had been a pre-med student in college. I figured my pre-med degree helped, but it didn’t really. The questions weren’t about the medicine, it was more about the conversation. The diagnosis were usually a sentence or two with a few medical terms that might have been unfamiliar to a layperson. I mean, I knew the physiologic abnormality they were talking about, but I didn’t know what to look for clinically or anything. <br><br>Anyway, with my first paycheck, I was able to get a better apartment. I still have no contact with the outside world, but I feel more comfortable. The one I was put in was some small thing with nothing in it but a bed. I had a break room and a workout room, but they were in the same building as my station. That was my entire world for the first month.<br><br>After that, I started questioning my situation a little more. Now I’m really questioning it a lot. I feel like something very wrong is happening.<br><br>A couple weeks ago, I had a call with my superiors at my station. Someone put a video conferencing display in the corner of the room and gave me instructions for how to call them, and once I did, they immediately picked up. It was two people in a conference room, both dressed professionally. I spent the first two minutes of the call trying to ask questions about what it was like outside, if I could contact my friends and colleagues, if I could go home, and they ignored me. Eventually, they made it clear that they weren’t going to give me any information I didn’t already have about the outside, and I should just focus on what they were asking me. I thought I was going to lose my job, so I shut up and did it.<br><br>After about two minutes, they told me that they wanted to test a new program, and they’d be sending me a couple audio samples in the next few days. The first one came about two days ago. I still had my normal audio samples from the last two days, but I got an email from an unknown email address (all my emails about my workday and other things came through the ClearVoice email system). <br><br>The email asked me to download the audio file and listen to it, then answer the questions in the email body. They said it was a conversation in a clinical setting, and I should be able to do it with my training. <br><br>When I listened to it, I heard a computer program talking to a girl about knee injuries. Apparently a kid had sprained/sore knee, and was using the voice AI, or voice assistant, or whatever to figure out what to do. She was talking through the program options for identifying and diagnosing what was wrong with her knee. <br><br>The questions were a lot different from any I had been given before. Most of them were things like what identifying questions did the program ask, and what were the diagnosis options that the program gave, and what was the recommended course of action. But the last question was asking about the little girl. How old was she? What was her symptoms? What was her diagnosis? What were the program’s instructions for the little girl?<br><br>I felt really bad. I wrote a long email back listing out the answers, but also asking what the hell this was for. They didn’t answer any of my questions, just said thanks. <br><br>I got another audio sample today. I was supposed to be done with my work for the day, because my next shift was going to be in 10 hours, but I got an email again for another sample from a voice AI. I didn’t listen to it yet. I just wrote a long email back about what the hell they were doing with this information. <br><br>I’m scared. This is definitely not for training an AI. It sounds like it’s training an AI to diagnose patients. I’ve been done with my shift for a couple hours. I think someone from management is going to come check on me because I haven’t played the audio sample yet.