In Horoma's Head 17

Hi everyone,

Month kind of flew by it seems. For this month's newsletter I want to do things slightly differently and expand on one specific topic rather than just throw several short topics. 
It won't even be much related to archviz, but more architecture and cities in general. But bear with me 😁

🐏🐏🐏

The world of Podcast (again!) 🕹🖥📊

This was scheduled for quite a while, but it finally happened and is online.

I've been invited by Georgi of The Creative Insider podcast to talk about me, my work, my career and plans for the future.

As usual even though talking about me is a bit boring, I like having conversations with other creatives in order to get other perspectives on some topic, and get some ideas out there for people to glean for there own potential use. 

The podcast covers many topics, from my year in Paris working in-house, to how I run my business, find client, how I work on my course, and other things like routine and where to get inspiration.

Hopefully you'll find the time to put it on while working and have me rambling in the background and get some useful insight. Feel free to get back to me if you want to discuss any topic if you feel like it!

The World of series and Japan 🕘🦺👓

I recently finished the series set in old Japan called Tokyo Vice. On top of having a great storyline, the photography is just superb. A lot of the japanese vibe is visible in many different type of shots with great moods and colors. I basically spent half the show taking printscreen for my own references folder.

The World of Dall-E and its competitors 🖥🎭🎨

I talked about Dall-E in a previous installment of this newsletter, and as for any promising market, there are competitors. Google's Imagen recently released some tests and Benjamin Hilton posted a thread on Twitter doing side-by-side comparison of both AI engines generating images from natural language descriptions. 

It's interesting to see how sometimes the result is very close and how sometimes one engine will be way smarter. 

I'm particularly thinking of the example of the "giant cobra snake on a farm. the snake is made out of corn" example where Imagen actually understood how to make the cobra out of corn where Dall-E simply put corn next to the cobra. Pretty mind blowing

The world of fashion and Kahn 👄🤟🎭

A bit over a week ago was published a great video showcasing the latest Louis Vuitton's fashion even that was hosted in nothing less than the Salk Institute designed by Louis Kahn. 

Don't pay attention to the horrible French accent (I doubt you can, but try), and just enjoy the nice colors and views. Also the clothing collection itself has some very interesting pieces with crazy fabrics that would be fun to recreate in 3D hehe... one day maybe.

The World of AI in Architecture 🥼👔🏢

Now onto me rambling about AI and architecture. I'm by no mean an expert in any of the field, but I do have this tendency of creatively connect random dots and wanted to share some things that crossed my mind lately. 

A bit of context

So, a couple of weeks ago I accompanied a friend to a lecture by Neil Leach at the IAAC on the topic of AI progress in the field of architecture. While I had an already decent knowledge of most of the topics covered in the lecture having them laid down by Neil in a coherent whole shed some light on how this topic can quickly branch out to many topics that are quite dear to me (whether it's ethics, the future of automation in architecture, urban planning, automation for image crafting, and more broadly what makes a city liveable).

Neil purposefully took a strong stance on the topic, foreseeing the death of architects (at least as we know it), addressing one of the most common objection architects have "our work is too creative for an AI or algorithm to do it". Spoiler alert, he actually is about to release one book covering the his pessimistic view of the field with the democratization of AI, and another book covering a more optimistic look. I personally can't wait to read them to broaden my perspective on the subject. 

Back to the topic of creativity though, I think there's much to be said here. I'll keep it short for this newsletter and just want to give you food for thought as I think it's a fundamental topic for the decades to come. 

Whether architects are creative or not, for now AI is indeed not capable of doing creative work as what it does is learning and generating output (wheter images, projects and whatnot) that are just further iterations based on what it learnt.

Creativity on the other hand is the capacity to potentially generate paradigm shift that completely bends the rules of what was done before. To my current understanding, this can't be done by an AI since it will always stay within the boundaries of what it's learning.

If I were to take a quick analogy, it's as if we were expecting an AI to generate the concept of a car after seeing many images of horses. I doubt it's possible, while this is the type of disruptive innovations that humans came up with more than a century ago. 

Now, are architects really that creative? After writing my thesis on the topic of urban regulations and its impact on architeture diversity, let's say that creativity will be confined to what is legally doable... which sometimes is not much. And it can therefore become more a question of exploring what is possible ... and when it comes to iterating, AI will always be more efficient than humans (there are some examples of those in the talk).

One aspect though that I think is particularly interesting to ponder, is that an AI won't be able to actually choose "what is best" according to humanistic criterias, because those criterias are biased and based on heuristics that can't really be reproduced accurately by an AI because they can't really be stated properly. An AI, on the contrary, will choose the best design according to the regulations and the brief, or at least how they were entered by the designer team. While this can seem like an ideal scenario, the paradox I'm foreseeing, is that we as humans are not used to living and evolving in "perfect spaces". There is always a weird detail, a little egotistic signature from the designer, or a simple error, that even though it means the building is not "ideal", it makes it more human.

The example that came to my mind after thinking about this aspect is the love Europeans have for organically (anarchically I could even say) evolving historic centres against more centrally planned urban fabric. While the work of Jane Jacobs or Christopher Alexander give us some pointers of how the urban fabric and relationships are built, the very essence of those places come from the constant error correcting that takes place when things get built next one to another, or on top, etc. 

Those buildings have absolutely nothing "ideal" from an objective perspective, but are very natural candidates for the love of the inhabitants. Can an AI generate that type of more organic urbanism without becoming either a pastiche of what we used to do several centuries ago, or an aseptized modern version of it that would feel nothing like the original to humans? And yet look like the most optimized solution according to the AI generating it.

Just some ideas that came to my mind after this short, yet fascinating, dive into the realm of AI. I might completely shift perspective in a week, or with further addition from info I'll gather over the coming weeks or months. I unfortunately didn't have time to discuss much further these topics with Neal even though we enjoyed some absinth drinks together with some of his students after the lecture. As usual, any tool can create good or bad outputs, and I tend to go on the optimistic side all the time. But I liked that this new potential paradigm shift can question more deeply the type of urban fabric we want to live in, and who can actually create it coherently and genuinely.

That'll be all for this month. 

Next month will normally see me finishing recording everything for Composition From The Ground Up. Let's see if that really happens :) 

If you find this utterly boring and useless, please let me know before unsubscribing 🎀. If on the other hand you do like it, then I'm glad, and feel free to share it with your archviz pals that would maybe like it too by sending them here : https://www.horoma.net/newsletter-subscription

In the meantime, all the best for next month, summer is nigh...

Cheers

Simon