AI, GPT-4, etc… The past week was so eventful that it felt like a year had elapsed.
During my spring break, I spent a day exploring Toronto’s Evergreen Brick Works, a place rich in history, nature, and community. The bricks produced here, utilizing the abundant clay resources in Southern Ontario, once shaped Toronto’s architectural landscape, contributing to iconic structures like Casa Loma, Osgoode Hall, Massey Hall, and the Ontario Legislature. As I meandered among the remnants of brick-making machinery and kilns, my thoughts gravitated towards the AI advances unfolding in our world. The brick factory’s century-old history echoes our current reality, urging us to re-think the technologies today.
The past week was so eventful that it felt like a year had elapsed. On Tuesday, OpenAI announced GPT-4, a chatbot boasting a vast array of updates over its three-month-old predecessor, ChatGPT, and equipped with astounding capabilities. Chatbots today can not only handle everyday conversations but also perform complex reasoning and mental arithmetic. They can pass major standardized exams, including the Bar Exam, LSAT, and GRE, often scoring higher than the average human test-taker. They can even do simple coding, a skill that many people dread learning, and troubleshoot their own bugs by interacting with error messages. GPT-4’s most impressive feature, compared to its predecessor, is its multimodal capability—it can process input in the form of images, essentially allowing an AI to “open its eyes.”
Within the same week, both Google and Anthropic—a start-up focusing on AI safety and research—announced the release of their own Large Language Models (LLMs), a general term for describing the AI technology behind today’s chatbots. Simultaneously, Microsoft showcased a future in which its Office Suite integrated AI assistants supported by LLMs. Meanwhile, Midjourney, a start-up specializing in AI-generated images, released its latest V5 engine, featuring photorealistic effects often indistinguishable from real images by human observers.
How do we comprehend the impact of these rapidly advancing technologies? Imagine being an individual content creator today—a writer, photographer, graphic designer, YouTuber, or TikToker. You would likely employ AI to unleash your productivity and creativity. AI can draft or polish your essay, create an illustration or video cover, stylize your photos, design a logo for your brand, and transcribe and translate your content for a global audience. For the self-employed, AI may now automate spreadsheets, customize customer emails, prepare tax forms, consult laws and regulations, and even make entrepreneurial decisions. AI has the potential to democratize most professional services, allowing everyone to fully embrace creativity and individuality.
It is tempting to compare today’s AI to the machines of the industrial revolution, as both elevate our productivity to new heights. Indeed, many people feared losing their jobs due to the diminishing demand for manual labor, much like in the old days. However, I believe the real cause for concern is the inexplicable, the je ne sais quoi—we don’t know how to explain our own creative processes, and suddenly AI can automate our innermost workings. In other words, the key difference between now and the industrial revolution is that we are reaching a ceiling in understanding ourselves.
Another distinction is our limited reciprocal understanding of AI, unlike the motors and gears we design. The emergence of capabilities in LLMs—a phenomenon where certain reasoning abilities appear in large models with over 10 billion parameters but are nonexistent in smaller models—remains one of the biggest mysteries for AI researchers today.
However, as I stood amidst the brick-making machinery, I realized that our journey to comprehend LLMs is not unlike the evolution of brickmaking, or even our understanding of life and mind. Humans began using sun-dried mud bricks before 7500 BCE in Mesopotamia, and fired bricks emerged around 4400 BCE in Neolithic China. It wasn’t until the 19th and 20th centuries that we gradually unraveled the science behind materials and high-temperature firing techniques.
Similarly, our understanding of life has progressed over time. We once believed in a vital force that animated living beings but eventually discovered that life is simply a complex arrangement of chemical elements. Our understanding of the mind has also evolved. We used to think of the mind as an immaterial entity, the soul, but now we tend to believe that the mind arises from the billions of neurons firing in our brains. In a way, LLMs could be viewed as a model of the mind emerging from simpler computational elements. As our understanding of the world is ever-evolving, the magic of today may become the science of tomorrow.
The Brick Works site has undergone its own transformation. Once a bustling industrial center, it was eventually shut down when clay resources were depleted, and the site lay abandoned for years, becoming a haven for skateboarders and graffiti artists. In the 2000s, a non-profit organization named Evergreen stepped in, renovating the site, filling the pit, planting trees, and preserving its heritage. Today, it serves as a vibrant community center, complete with a children’s playground, local farmer’s market, and energy-saving buildings that house galleries and greenhouses.
As we continue to witness the unfolding AI revolution, I wonder what legacy we will leave for future generations. Perhaps, the unparalleled surge in productivity may lead us to an era of boom and bust—numerous ventures fail while only a few prevail. The surviving AI entities may gain power and redefine the global order. Or perhaps, the widespread dissemination of creative capabilities could liberate individuals from traditional work, enabling them to embrace newfound identities. Historically, economic progress has often led to the healing of societal wounds and the pursuit of harmony with nature and our heritage, even as we push the boundaries of innovation. Standing at the brink of this transformative singularity, we collectively navigate the uncharted waters of a social metamorphosis.
What an exciting time to be alive.