AI and the Unforeseen Ripple Effects

Doom. What could come of it? It is not the post-apocalyptic future but rather how sometimes the beginning of something can lead us to places we do not always anticipate.

In 1993 DOOM the video game, was introduced and became one of the most popular first-person adventure games.  In part, it was famous for its rendering of textured background graphics and even allowed users to create their own landscapes. As a 3D graphics game the computational requirements were extreme and as a result the game along with its successors ushered in the development of something completely new – the GPU or graphics processing unit which is a computer chip solely dedicated to doing the high-end math required for dynamic graphics.

Of course, the game developers did not dream up the GPU, they only created the demand. Instead, a trio of smart engineers saw the convergence of popular video gaming and the rapid advancements in computing power and founded NVIDIA that same year.  The competition in video game consoles brought new investments and drove innovation into something that might not otherwise exist. NVIDIA as we now know, is not just a company that makes GPUs but those chips and their high computational power spawned in the 1990s are also the key technology needed for Generative AI and are behind many of the ChatGPT-type tools.

The Adjacent Possible

Would ChatGPT, Claude, Perplexity, and other large language model transformers exist without NVIDIA and the GPU? Likely they would but their rate of development and capabilities might not. By the time eight Google engineers published their landmark whitepaper, Attention is All You Need, outlining how a transformer model could work in 2017 the GPU was ready.  From the mid-1990s to 2017 GPUs became 25 times more efficient and as a single-purpose device outpaced development in a computer’s main processing chip (CPU). Video gaming helped to create GPUs and GPUs helped create Generative AI. Without these adjacent technologies, where would we be? And how much longer would it take to make great ideas happen?

Power of AI

Generative AI and its most popular tool OpenAI’s ChatGPT have only been available for two years. In this time we have seen several improvements in its capability and more importantly a wide range of ideas of how it can best be utilized. From healthcare to graphic design, it is safe to say most industries and the way we all do work will change.  If you are skeptical consider that in the U.S. several of the big technology companies, Amazon, Google, and Microsoft, are reshaping the power grid.  All those data centers filled with GPUs running AI need a lot of electricity and they are turning to nuclear-powered electricity, an industry that has previously stalled in the U.S. having only built one new reactor in the past two decades. With AI things are changing. Consider these recent facts:

• Microsoft has a 20-year agreement and will spend $1.6B to restart the Three Mile Island Nuclear electric facility in the state of Pennsylvania.
• Google is partnering with Kairos Power to provide smaller modular nuclear reactors by 2030, each costing $1B, and will co-locate them at data centers.
• Startup TerraPower is building a new approach to nuclear-powered energy, also sponsored by Microsoft former CEO Bill Gates. Their first reactor will be in production by 2030.

---

AI Butterfly Effect

AI, as a transformative technology, has significantly advanced our understanding of diseases and enhanced creative applications such as AI-enhanced photos. These innovations build upon the foundational ideas of the original Google engineers and demonstrate the boundless potential of AI. However, as we look to the future, it becomes clear that we are only beginning to understand what AI can achieve and influence. The true extent of its impact remains uncertain, much like the unforeseen butterfly effect the video gaming industry of the 1990s had on other technology developments. As AI continues to evolve, it is likely to create new demands, opportunities, and challenges that we cannot yet fully anticipate.  Ultimately, the future of AI is a complex tapestry of possibilities, and it is up to us to navigate and shape its trajectory with careful consideration and foresight.