As AI technology becomes commonplace in many industries around the world, new applications are constantly being found.
Web design and building has evolved rapidly in the past two decades, with organisations now able to craft vibrant and intelligent online hubs for whatever they choose. Wix has been at the forefront of this revolution, and is looking to closely combine AI and website building - we spoke to Nitzan Achsaf from Wix to find out more..
- Here's our round-up of the best website builder solutions of 2018
How can you have an AI design product that enables quick creation but a unique output?
Quick creation is done since the machine learning models are already built, so you just need to run them on user inputs, and that's instantaneously. Unique output happen since the amount of permutations, between business types, text, fonts, colors, layouts, section types, number of section on page, and more are enormous.
It’s about creating something that can evolve over time. From the beginning of its development, the information that’s fed into it should be high quality, diverse and continue to evolve. It doesn’t end when the product is released - continuously it must be updated. Also, see what is created when the product is public and the user gets their hands on it. What changes are made and how that could be a preferable outcome.
How does an AI tool keep up with new functions that are added?
Today, websites are more than just business cards - they’re the central point of running a business. An AI design product should be built towards that. The AI knows which designs or features work better than others. For instance, for converting visitors into leads. This also helps the AI to rank the relevant results for the user.
Look at what users are adding afterwards. If websites created for photographers routinely have calendars added to them, then hopefully that information is making it back into the system. So when the next photographer comes looking for a website, the calendar is built in. These tools should learn to be anticipating more than just the good design but what features, sections and pages they need.
How does an AI product keep up with recent design trends - especially when they can change quickly?
Users can experiment with creating many different sites, as well as designers bring new trends to the market. Over time the AI needs to identify new layouts and designs that are coming out and know when and how to add them.
I would say that the hope is an AI-lead design product would start creating design trends. Like I said earlier, it’s important to keep it learning - from its own outputs and the outside world - and to innovate from what it is pulling in. Also, if a trend is more color based then the output should be easy for any user to change - not just a professional designer. If the ultimate owner a website wishes to move from pale pink to light grey, it should be easy to accomplish.
Is there a worry that sites will be created but hard to manage?
It depends what the AI is trying to optimize. For instance, If the AI knows how to take into account layouts that don't overlap with each other for easier editing, it shouldn't be a problem. If the product isn’t thoughtful, then sure. But it’s important to create user friendly and intuitive backends that allow for easy management. These websites are more than brochures. They must be easy to update and add text, change simpler design parts (like color, images or font), or even duplicate pages.
How can a professional designer use AI?
The way they use any tool. One way that they can use it is to create fast drafts to their customers before the final design. Another way it to get new ideas before they start their own design work from scratch. Use it to build a final product for themselves or a client. The output can be the start and hopefully these systems allow for more control afterwards. Of course, whoever is managing content will need to be able to add and update what’s necessary.
What does AI design tools mean for the content of the website?
As a system can learn what features and functions a website could use, it should use natural language processing to learn the content. That could be from scanning a previously created content, learning what each block of text is and placing it automatically in the new site or by suggesting text (a smarter lorem ipsum). We’re at a point where a system could move text from one site to another but not yet where the content is generated uniquely and won’t need to be adjusted all.
Where do you see it going next?
As AI progresses, it will be harder and harder to know which site was built via AI and which was built via a designer. You can think of it like passing a "design turing test", i.e. in the future humans will not be able to differential between the two. Then, it’ll have to get innovative. Instead of mimicking what it is learning from what’s created, it will get better and more experimental. It’s easy to see how soon most websites that are created use AI in some way.
Nitzan Achsaf is VP and GM of consumer experience at Wix.
