When Stable Diffusion started being more accessible to casual coders like me back in 2023, I immediately saw the parallels between that technology and some of the sci-fi tech I grew up with. Typing up a prompt to then have the generative AI spit out a relevant image wasn’t much different to me than the way the crew of the Enterprise would use the holodeck.
In the episode “”11001001” of Star Trek: The Next Generation, Riker says, “Computer – I need a place to play some music – a little atmosphere.” before stepping into a holodeck transformed to suit his needs. He even cleans up his prompt by specifying that he wants the era to be “Circa 1958,” and refines it even more by saying, “Kansas City. No, wait. New Orleans. Yeah. New Orleans — the Low Note. ‘Round midnight”. Once he steps into the jazz bar and marvels at the computer’s work, Riker has it add more things to the room, such as a band and an audience. What the computer is doing is generating content based on Riker’s prompts.
It’s exciting to think that we are living in an age where such foundational, transformative technology is available to us. We stand on the precipice of an Intelligence Revolution that will exponentially increase humankind’s computational ability. We can model complex data and mathematical problems with a simple phrase. We can create engaging images and videos to communicate ideas that we couldn’t articulate to others very well without extensive art training. We can run simple experiments millions of times digitally and observe novel results that would’ve taken us decades to realize in the physical world. The possibilities of this technology seem endless, but its issues are also myriad.
Holodecks are self-contained systems aboard starships and other high-tech facilities. They tend to be about the size of a large living room, though they possess the ability to create environments that perceptually extend vast distances, like entire cities or oceans. Our modern AI technology is a lot more sprawling. Just as early computers were building-sized, so too are the vast data infrastructures that enable AI companies like xAI and OpenAI. These infrastructures span across nations, providing cloud computing as well as a wealth of data storage. Most technology reduces in size as it evolves, a process known as miniaturization. However, the physical structure of AI data infrastructure is expanding at an exponential rate all over the world, consuming more and more resources in the process.
There are the environmental concerns of this tech, such as how entire small towns are being converted into data centers and heat sinks. There are also the issues relating to how this technology is owned by billionaires who already have an outsized amount of power and influence on society. Millions of people are feeding these AI models personal information at a speed unheard of even throughout our social media age, further concentrating vast swathes of sensitive data into the hands of psychopaths and megalomaniacs. And of course there’s the unprecedented amount of digital theft that birthed these generative models, which seemingly used the entire internet to train on.
Despite the ethical challenges of this new technology, generative AI is here to stay. It has brute-forced its way into our society, as has already begun to reshape it. There is no putting the genie back in the bottle, regardless of how much fear and actual harm that it is doing. When there is this much money behind something, its motive force becomes unstoppable, at least in the short term. Many of the innovations that have shaped contemporary civilization have had messy beginnings, such as the tremendous pollution that plastics and other petroleum products have done. Humans have a habit of widely adapting technologies that make their lives easier or more entertaining, and capitalism turbo-charges the availability of those technologies.
I’ve always wondered how humanity gets to these space-faring futures, like in Star Trek. There are many “great men” theories and portrayals in fiction, such as Zefram Cochrane creating warp drive. However, most transformative scientific progress has been slow and collaborative, building upon the knowledge accumulated by humanity over centuries. A few people might make leaps in logic, but those observations are also informed by collaboration with experts in their fields. There would be no Einstein without Grossmann and Hilbert, no Newton without Leibniz and Halley.
Generative AI, when it is less error-prone, offers the ability to collaborate with vast bodies of research and historical knowledge. In this way, it can accelerate our sciences and serve as a quantum leap towards a better, brighter future. However, in its current form, AI technology just isn’t robust or accurate enough to provide this level of scientific collaboration outside of highly-trained, project-specific, and often proprietary models. Imagine being able to work alongside the greatest minds of our era, or creating composite characters of those minds, just as Janeway worked with Leonardo da Vinci within Voyager‘s holodeck.
The holodeck offers a way to turn one’s imagination into something that they can interact with in the real world. Generative AI systems like Stable Diffusion and Gemini can already create images and even videos tailored by our imaginations. The holodeck computer is perhaps the culmination of the generative AI models we are creating today, refined and perfected. Heck, even the replicators would need to use such generative AI to create dishes on the fly with limited audio prompts. Now we just need the antimatter or fusion energy sources, the photonic emitters, the force fields, the matter reconstituters…
Disciple of the Vault 