Systems Final Project and Overall Impressions
Systems Thinking Wrap-Up
One more semester done! This week my final project was due for Systems Thinking. The class has so much good content, and it was great to be able to have someone of the caliber of Wendy Schultz there to answer our questions from her vast experience in the field of futures. I do wish we had spent more time directly applying the work to futures projects, and an overview and pointers to quantitative work in systems like simulation (how it’s used in futures, the key approaches and people/groups, etc) would have been helpful, but overall I’m glad this is part of the degree and I loved interacting with the other students and getting to know them better. My next class will be Social Change in the fall, so I’ll be taking another summer off from formal study and this newsletter will again shift to looking at the future through books, film, and other interesting things I’ve been collecting in a list but too busy to dig into during the semester.
Here’s my final project, about using systems thinking to better understand the current state and potential future outcomes of the AI industry. It’s an example where the future is coming at us pretty fast, as the OpenAI GPT-4o demos earlier this week demonstrated1. Honestly I wish the project were a little more in-depth; I feel like what I covered was kind of superficial, but the topic is so huge it was hard to do otherwise. If I had committed to a topic earlier and focused all my assignments on AI2, I would have had a much deeper well to draw from. FYI, if you’re reading this via email, you may only see a few of these images without clicking on an “expand” button or clicking through to the web version.
This demo (and my experience playing with it) was fascinating because, although the intelligence isn't much higher (suggesting we may be close to the end of what can be learned via text alone), the native multi-modality and greatly reduced latency feels like as much of an experience upgrade as the initial ChatGPT was from the LLMs that preceded it. Add to this the fact that OpenAI is changing their business model to give free users the best model (but lower volumes), and there should be another big jump in utilization in coming months.
Originally I wanted to look at the future of trust, and how AI and other social changes affects our ability to trust the things we read online, photos and video, etc. However, trust is so amorphous that it’s hard to draw boundaries around it; it’s much better as a variable of interest in a system than as the domain/topic of the system.