It often seems like there’s a general consensus that human content is “real” content, with AI generated content being a sort of derivative “copying” what humans do.
Notwithstanding the rapid transformation of simple GAN network models to vibrant generative AI, we still want to see humans as being the better creators. But that may not be the case for long, and that consensus may, in some ways, be crumbling.
I took a look at this recent X post by Aidan McLaughlin of Mediaite fame, who indicated that he’d prefer to have the computers designing buildings.
“The proliferation of fun, maybe impossible, surprisingly tasteful, and deeply calming AI architecture is so cool to me,” McLaughlin writes. “AI should commoditize beauty. I think there’s a happy chance stuff like this raises my generation’s standards, and my children grow up with less human slop.”
Human slop?
This contrasts with so many comments talking about how the products of LLMs are only based on the human content they mirror, and how there needs to be human content at the core of training systems. Many analysts have suggested that without the original human content, AI becomes a scavenger of its own products, a recursive rehasher that ultimately provides a watered-down gruel of bloodless stuff.
In terms of architecture, the examples in McLaughlin‘s post look good – a warm set of living spaces, with hot colors offering a respite from a snowy landscape – and everything looks very well put together, but are the AI results “better?”
Architecture, of course, is one of those realms where quality is pretty subjective, and taste (along with utility) is the primary guide, but there’s also McLaughlin’s characterization of AI designs as “maybe impossible,” meaning that they haven’t been vetted completely according to the laws of physics, and that an AI architect is prone to some hallucinations as it designs. As for “deeply calming,” many people find generative AI products to be alluring, seductive, tranquil, or otherwise attractive, but just as many are sort of nervous when perusing these kinds of images, at least partly because they have concerns about how genAI fits into our world.
But then there’s McLaughlin’s suggestion that AI designs are just nicer than those penned by human hands. To be sure, many of us talk about AI visuals as being too flawless, or “nice,” as a kind of drawback, but what if that ends up being what we prefer?
What’s in a Game?
For a good example of AI leading the way in design, let’s look at the digital spaces that people play in.
GenAI has become so good at creating its own game content that we’ve developed a term for it: procedural content creation – where game environments are created with algorithms, instead of manually by humans.
The result is endless worlds – in Minecraft, Roblox games or some other kind of digital construct. The Metaverse, it seems, will be largely designed by artificial intelligence creators. Will they have names?
The groundwork has already been set: the rise of no-code in platforms like Roblox dates back a couple of years.
“Earlier this year, Roblox said it was testing a tool that could accelerate the process of building and altering in-game objects by getting AI to write the code,” reported Marcus Law in AI Magazine in 2023. “The tool lets anyone playing Roblox create items such as buildings, terrain, and avatars; change the appearance and behavior of those things; and give them new interactive properties by typing what they want to achieve in natural language, rather than complex code.”
These ideas are being implemented in bold new ways.
“PCG provides technical artists, designers, and programmers with the ability to build fast, iterative tools and content of any complexity, ranging from Asset utilities, such as buildings or biome generation, up to entire worlds,” writes a spokesperson at Unreal Engine, describing those assistive processes.
How long will it be until AI erases the human in the loop?
Tools, Architecture and Architects
In a google search on architect job displacement, a spate of articles promises that “AI will not replace architects” though many of these writers admit that lower-level tasks will be automated, diminishing workforce demand. Then there are the actual architects, reporting from the trenches, for instance, that they are asked to use something like Midjourney for renderings. Check out the tool itself and what it can do to make words into visuals.
It’s not hard to find testimony from architects talking about how they use AI, either. A Reddit thread gives particular insights into how the tools are working as assistive support for human developers.
All of this illustrates how the technology is quickly revolutionizing this one part of the business world.
The Hardware Revolution
As companies gear up to take advantage of these new capabilities, the hardware industry is turning at a volatile pace…
I’ve been writing about the rise of Nvidia and other chip makers, and about TSMC‘s supply chain realities, and how countries like the U.S. and China are jousting for geopolitical positioning in 2025.
Simply put, the capabilities themselves are going to create enormous demand. Whether AI actually designs “better” than humans is a separate question – but the world is going to be observing a change that’s unprecedented in all of human history – for the first time, there are entities, albeit digital ones, that are in many ways smarter than us. As humans, we were smart enough to take ourselves almost completely out of the food chain. AI entities are even further removed from those biological processes that involve mortality. If they can live forever, how smart can they get?
I’ll leave you with that question, as all of us continue to react to these kinds of trail-blazing advancements in the new year.
