In the beginning web developers created sites using HTML. Then the internet was formless and empty, darkness was over the surface of the deep web, and the Spirit of Code was hovering over the water.
We’ve come a long way since then. The internet is still a dark, dreadful place, but it’s much more stylish now. We have web design to thank for that, which has grown exponentially in scale and sophistication over the last 25 years.
In fact, it’s gotten so sophisticated that we’ve started building artificial intelligence to do it for us. The technology is still in its infancy, but machine learning is enabling artificial design intelligence (ADI) to understand creative rules and apply them independently. Artificial design technology will soon be advanced enough to automate a lot of web design work. There will be thought leaders, and there will be their digital disciples.
People are slowly but surely waking up to the potential of the technology. Talk of automation tends to mean a lot of sweaty palms, but open arms is the best policy. We may be on the cusp of the ideal design assistant.
What is Artificial Design Intelligence?
Before we get ahead of ourselves, we should probably review what ADI is and where it’s at. Put simply, it’s artificial intelligence that uses machine learning to identify and implement web design trends. Put simpler still, it’s technology that can make websites on its own. Users state what they want and the ADI generates a personalized design.
That’s the idea. Its development has only taken hold in the last few years. Dedicated platforms like Bookmark and The Grid have cropped up with in-built design assistant technology, with the latter raising $4.6 million back in 2014. Meanwhile, established players in website building have taken similar strides. Wix released its ADI interface in mid-2016 and has been quietly developing it ever since, while GoDaddy has started to dabble in website generation.
To give you a sense of the technology’s growth, here are just a handful of the platforms exploring artificial design intelligence (to varying degrees):
What are they each trying to achieve? At the moment, if you want a website you either do it yourself — from the ground up or in website builders — or get a professional to do it for you. The former requires time, patience, and a degree of design savviness. The other tends to require quite a bit of moolah.
Artificial design intelligence, as Chitraparna Sinha puts it, is an attempt at a “third way”, a means of building a website without hiring a professional or doing it yourself. Nitzan Achsaf, the head of Wix’s ADI project, summarises neatly: “It’s trying to take the brain of a designer and make it into a machine.” How exciting, and slightly ominous.
This is done by making the programs capable of understanding design principles and then ‘observing’ what existing web pages are doing. Collating the two produces a universal average of web design. Theoretically, it cuts out the need even for templates. ADI can produce functional, attractive sites completely independent of human expertise.
Eves and Frankensteins
If this sounds too good to be true, that’s because it is. The technology is yet to catch up with the idea, and idea itself is not beyond criticism. ‘Automated creativity’ is a slightly sinister oxymoron after all, and the beautiful dream can get away from itself.
The Grid hasn’t exactly covered itself in glory since raking in those 2014 investments. Its early efforts were gleefully (and not unfairly) mocked by totally-not-nervous web designers when they surfaced in 2016 — not the best time for the Grid crew to be on a “team-cation” in Hawaii. Today The Grid’s stagnant online presence, role-modelled splendidly by its dormant social media accounts, has the resounding sense of a company that has slunk away and is waiting to be forgotten.
Even Wix’s ADI has kept to itself in the years since its release. Intelligent design sounds great, but when the results have more in common with Frankenstein’s monster than Adam and Eve they’re probably best swept under the rug.
Augmentation or Replacements?
Getting the most out of ADI means playing to its limitations as well as its strengths. Automation in design is nothing new, and if embraced it can free up time for the task most deserving of our human touch. As Yury Vetrov, Design Director at Mail.Ru, puts it, the technology offers an opportunity for delegation. “Designers would become art directors to their new apprentices, computers.”
A lot of design is not creative. A lot of design is rules and best practice, things that are teachable… programmable. This is probably cause for bad human designers to be worried, but their competent counterparts will ultimately be the ones dictating how ADI behaves and evolves. Silka Miesnieks, Head of Emerging Design at Adobe, summarizes nicely:
In most cases, AI only affects a percentage of a job a person does, but rarely the whole job. And if you free up workers to do more of the creative and other work they are good at, while leaving other tasks for AI, you can grow your organization.
Webflow and Bookmark are great examples of platforms committed to the ‘assistant’ bit of design assistance. They’re targeted at two very different skill levels, but the underlying function remains the same. Vlag Magdalin, CEO of Webflow, hits the nail on the head. Better design software isn’t about automating design, it’s about “automating away the repetitive programming bits as much as possible.”
And what about non-designers? That bassline competence is all a lot of people need. Humans are capable of very bad design, while the learning processes of ADI ensure, at least in theory, a polished, semi-professional result every time. It may not be unique, but that’s not always necessary. A lot of websites just need to work.
This is currently where platforms like Bookmark operate. Their virtual design assistant, AiDA, sits within the website builder itself. When users need help or advice — be it for styling or search engine optimization — AiDA walks them through the process and makes suggestions.
The Creative Leap
This is the impasse ADI finds itself at. Is it destined always to be a reactive, Igor-like force, or is there scope for it to bring its own form of creativity to the table? When approached for comment, Bookmark CEO David Kosmayer kindly agreed to offer his insights into the industry and where it’s heading.
This is where the technology can play to its strengths. AI can track user engagement with websites, and so ADI can potentially notice and execute design opportunities that humans can’t see. Bookmark is working on a two-phase plan to this end, the first of which revolves around “tracking and adjusting website sections based on the highest user engagement.” Phase two would offer “the ability for website owners to track incoming visitors to better tailor content on their website.”
Instead of shaping themselves around broad engagement, sites would present themselves differently to different individual users. For example, Kosmayer notes, “a male 18-24 would find different images and videos appealing than a female 25-35 would. Also, a person who lives in North America may be interested in different items than a visitor to your website from Europe.” For Bookmark, “The ultimate experience would be, after we ask permission from the website owner, for us to execute design changes on the fly for specific demographics.”
That extreme is perhaps a step too far for those concerned about global robot takeovers, but the direction makes sense. The happy compromise for Kosmayer is, “to have suggestions daily, weekly for customers telling them opportunities they are missing based upon our findings by examining their website compared to others.” The more sites in the system, the more sophisticated these suggestions become.
This ability to make creative leaps — to have and test eureka moments — seems like the current frontier of ADI. Does design intelligence need to make mistakes to work properly? Apparently not. It can use its engagement tracking ability to trial deviations and cater design to different users.
Adobe CEO Shantanu Narayen claims, “[N]o machine will be able to mimic the creative ability of the human mind.” Perhaps, but they can mimic the rules, and would it be so much of a leap to start subverting convention and testing the results? What is innovation if not breaking the rules and seeing what works?
It seems like ADI may be on the cusp of something special. To create something that applies programmed rules is one thing, to create something that can learn and apply new ones in real time is a whole other ball game.
British computer scientist Alan Turing, widely regarded as the father of artificial intelligence theory, laid out this principle all the way back in 1950:
Instead of trying to produce a programme (sic) to simulate the adult mind, why not rather try to produce one which simulates the child’s? If this were then subjected to an appropriate course of education one would obtain the adult brain.
As WordPress guru Chris Lema notes, real time learning is essential. Every new site is an opportunity for ADI to learn what people like, what they want, and what they think they want. What’s to stop it from recognizing and different rules and giving users the option to toggle? Have a site designed by a brutalist ADI, or minimalist ADI, or oddball maverick AI. The scope is huge.
A website that can adapt itself to different demographics and regions using engagement data is a fascinating prospect. If ADIs start to behave like the big robot in The Incredibles, then we’ll really be onto something. No city will be safe.
It’s easy to get ahead of oneself with potential technology. There will be hurdles. It requires unprecedented sophistication and adaptability. In design of all things, the first impression is key. A product that talks a good game but which is laughed out of all the forums that matter is not one destined for success.
Not to mention the possibility of the machine learning systems being abused. The internet shows no mercy to content-neutral AI algorithms. It’s a naive soul who expects to reach ADI perfection without having to endure some spectacular trolling campaigns.
Also, as David Kosmayer points out, it’s not in everyone’s interests for ADI to become too sophisticated. “Templates still rule the day for the majority of the industry,” and superb website generation makes website building unnecessary for a lot of people.
A customer base has been trained to use templates and a certain way of building websites, earnings and growth need to be maintained. Trying to advance how websites are built is not a priority for the largest companies.
This, presumably, is why big players like Wix are working on ADI in parallel with their core product, or taking the GoDaddy approach of barely evening acknowledging it’s in their product. The principle is there, and it’s being developed by startups and conglomerates alike. It will be interesting to see which brings about the tipping point.
Web designers are not going to become obsolete, but a lot of the work they do probably will. As ADI gets more sophisticated it will be able to provide a bassline level of competence which can be utilised by professionals and commoners alike. As it gets smarter still, well, then things will get really interesting.
The real question is, Who’s going to crack it first?