Collapsing My Vibe Function

Writing for myself and for anyone who finds my thoughts interesting

Photo by Billy Huynh on Unsplash

An interesting question is how to start a page such as this without knowing what an audience may look like. Is this writing simply for the void, the occasional curiosity of my friends and family, or someday thousands of people whom I will never meet? Given this uncertainty, I figure it is best to approach this as simply a personal endeavor to gather my thoughts concretely.

While I’ve written before for classes and other nonsense, rarely have I written about topics that are dear to me for the purpose of depositing my thoughts in text. So far, I’ve been satisfied with simply allowing my thoughts to exist as immaterial vibes, only extracted in the moment of reflection or conversation. With my deepening involvement within the AI safety space post-graduation, I feel that it is prudent to begin collecting my thoughts across the massive topic that is artificial intelligence and its implications for humanity to have a repository of my thinking to reference for myself and for others.

I view artificial intelligence as the inevitable result of the grand project of science and mathematics over the last few hundred years. Ultimately, despite my love of mathematics, I’m deeply concerned that we’re playing with increasingly hotter fire and eventually we’re getting burned… and it’s all (mostly) the fault of us math nerds.

Who am I?

I figure some introductions are in order if you’re unironically reading my introductory post which contains no real substance compared to the other posts I’m planning to write.

Hi, my name is Max Winga, I’m a 21-year-old physics graduate and AI safety researcher. My personal ethics flows from a place of wanting to improve the lives of others and ultimately make the world a better place with less suffering. This is why I’m dedicated to working on AI safety, because I believe that without immense global cooperation, we are probably witnessing the end of the story of the human race.

I’ll explore my beliefs around AI in dedicated posts on this page, and will end this introduction with a short autobiography. Structuring my life into a few paragraphs is an inherently reductive endeavor, but I’ll give it a shot anyway:

I grew up in a small town called Manitowoc, WI, located on the shore of Lake Michigan, and notable for being home to a WWII submarine, Netflix’s Making a Murderer, and not much else.

My early life was fairly isolated, going out for school and church, but otherwise only interacting with my parents and younger brother until around 8th grade. My parents gave me an incredible gift by sending me to Montessori education until 6th grade, which nurtured my insatiable curiosity and love of math and science. The use of physical “materials” to teach mathematical operations built the foundation of my geometric and intuitive understanding of mathematics.

I continued beyond Montessori to attend Catholic middle school and high school despite not being from a Catholic family (in fact, I had stopped believing in religion in short order after finding out first about the tooth fairy, then Santa, only to be surprised when people were serious about religion being real.) Here, after skipping a grade, I continued my voracious consumption of knowledge and excelled across the board in my classes, ultimately achieving Valedictorian among my admittedly small class of 50-some peers.

I’ve always had a love of video games and computers, starting from an early age with a few Wii games, and then advancing to Minecraft for much of my early years. Along with school, I worked a number of service jobs to fund my interests in PC gaming, which gave me some of the first real grounding in the world outside of my bubble.

In my junior year of high school, I took AP Physics which showed me for the first time real applications of all of the math I’d been learning, and I immediately fell in love. Despite no physics classes in my senior year, I was set on pursuing physics in college.

Going to college was always a given for me, eventually landing me at the University of Illinois Urbana-Champaign…right in the thick of the early COVID pandemic in 2020. While COVID neutering the college experience certainly sucked, college-with-COVID was still much more enjoyable than high school and restrictions slowly lifted throughout.

Despite the pandemic, I took to physics with gusto and loved learning about math and the universe through the physics lens. I also found great enjoyment out of coding through introductory courses and labwork in my early degree. Thus, the stage was set for my introduction to artificial intelligence in the summer of 2022.

That summer, I was working with a miserable old professor on a project to automate the counting of corn rootworm beetles in glue traps. Specifically, I was tasked with getting an Arduino with 10 Kb of RAM to run species-specific image processing on color images (a helpful netizen compared this to being asked to fell a redwood with a handsaw.) While the project collapsed because the professor stubbornly insisted on using the Arduino, I learned a lot about image processing, which taught me about neural networks for the first time (beyond some Code Bullet videos in high school.)

This sparked an interest that would soon explode into my primary focus with the release of Midjourney and ChatGPT in the coming months. These releases, combined with a deep dive into the field of machine learning and hours spent listening to voices from across the field discuss AI, made me realize the true scale of the implications of what was happening.

What was initially optimism for the soon-to-be was turned into deep concern by learning about our lack of understanding or control over these systems and the risks they will pose as we scale to super-human intelligence. When GPT-4 released over my junior year spring break, I realized how quickly this was happening and I was sent into an existential spiral worried about extinction.

I came out of this low point fully dedicated to spending my career focused on these problems. I added a CS minor to my degree which I used to take all of the AI courses at my university including two foundational courses, CS 442: Trustworthy Machine Learning with Han Zhao, and CS 562: Advanced Topics in Security, Privacy, and Machine Learning, my only graduate course, with Varun Chandrasekaran.

I recently graduated on schedule in May 2024 with highest honors and despite having accepted an offer of admission at UCSD for a Master’s in Computer Science, I’d been feeling fairly unsure of where I wanted to end up in the world. I don’t think we have much time left of a “normal” world at the current rate of AI advancement, and I worried that I would regret spending a decent chunk of that time twiddling my thumbs in grad school.

I began to work with PauseAI, an activist group with a wonderfully descriptive name, and then out of the blue, heard back from an application I had submitted months ago to Conjecture AI. After making it through the extensive interview process, I’ve accepted a role as a Research Engineer, and will be beginning in early September! My work will hopefully contribute to more controllable and safe AI systems that are an alternative to the massive black-box models that are currently being recklessly developed.

Conclusion

It’s actually wild that you just read through this entire post, in fact, it makes me somewhat question your time-management skills. My future posts should be more interesting than this as I’ll focus on specific topics and dig deeper into them. Some ideas I have in the works (in no particular order) are:

• My views on why we likely cannot control super-intelligent AI systems

• The ethics of building truly intelligent AI systems

• AI as the next stage of evolution

• Meta thoughts on having conversations about AI risk

• AI and robotics

• My views on consciousness in humans and AI

• Why I believe a pause on frontier AI development is necessary and how it might be achieved

• What AI “art” lacks and why human art will always be special

• Comparing AI to our conception of alien civilizations

• Why the tribalism of “short-term vs. long-term risks” is harmful to everyone

• Ok we actually don’t talk about the military enough. This is going to be really bad. How are we just letting militaries go 120% into automating warfare as fast as possible?

• …and other ideas that aren’t on the top of my head right now

I guess this is where I say if this sounds useful, subscribe or something to see these posts when they come out.

The world is insane right now, and it will only keep getting crazier. I’ll be riding these rapids along with you and everyone else in the hopes that my paddling can contribute to keeping us afloat.

I wish the best to all my fellow humans, good luck out there.

---