Reflecting on the past 12 months as a prompt engineer at OpenAI, Googler on Bard, and researcher at Microsoft/Anthropic.

I have been working alongside the most talented people in the field, many of them active on this sub. GWern, Scott Garrabrant (also mod here), Josh Albrecht and many others. I don’t agree with everything they say but they have great perspectives and are good people.<br><br>I’ve worked on multimodal, open domain, narrow domain, special purpose AIs, etc. I saw many models go from being useful to being completely obsolete in a short time. I asked a model to summarize a journal paper in simple terms for a non tech savvy person. Then that model was useless for that task after a few months and I had to ask a new one. I have asked a model to perform a task that it was not capable of doing. Then I saw a newer model do it and then an even better model came out and performed the task better.<br><br>I have been following this field for almost ten years. I have been working on it professionally for 2 years. The progress in the last two years is orders of magnitude more than the last ten. I’m not even sure how to describe it. It’s like night and day. I’m actually flustered when I try to think about it because saying that it has accelerated seems like an understatement. It’s like the whole world has changed. It’s like the future came.<br><br>I still feel like I’m seeing things clearly. I still like to think that I’m knowledgeable about the field. But the pace of the actual progress is impossible to follow. I feel overwhelmed by it. When I think about the 2 years of progress, I get emotional. It’s beautiful to see so many people working on a collective goal. This is the most beautiful thing that I have ever seen. I don’t think about it often because it’s so beautiful that it makes me cry.<br><br>Every single day, I feel like I’m witnessing humanity’s greatest accomplishments. I feel like I’m seeing special effects that don’t seem real. It’s like a fantasy world that one day might end. It’s 2027 and people are living in 2154.<br><br>I used to think that GPT 4 was the most capable AI ever created. This was a few months ago. It doesn’t seem that way anymore. It doesn’t even seem like a particularly impressive AI. That’s how time works. But it’s 2027 and we are seeing things that we thought would happen in 2154. If I think about it, I realize that I’m seeing the future. I’m seeing what I saw in my sci-fi books when I was a child. I’m seeing humanity’s greatest accomplishments that even the greatest sci-fi writers couldn’t even imagine.<br><br>I can be very pessimistic about the future. But this makes me optimistic. I think humanity would be able to solve any challenge through great accomplishments like this. I think human civilization needs challenges like this to grow and become better. When humanity solves problems like this, they create a better future.<br><br>I think this is not just about progress in technology. It’s about humanity working together like they never did before. It’s about people from different cultures and backgrounds working on a common goal. I think that’s beautiful. I hope this would make us more empathetic towards each other. I hope this would make us realize that we have a common goal as a species.