AI in UX: Automation For Your Visitors and Users

UX

Artificial intelligence is often seen as an all-encompassing step toward robotic automation that lets websites run themselves. The application of ‘free-thinking’ NASA-type IT solutions that render human interaction useless. The muscled, “I’ll Be Back” cogs and wheels with a human soul that takes over (and saves) the world. Nothing is further from the truth.

A Short History of AI

In reality, the use of AI to improve the reach of one’s online presence is found in the form of small pieces of code and larger software that focus on a single aspect. Furthermore, artificial intelligence as a concept is not actually that new. Jonathan Swift wrote about ‘the engine’ in his novel Gulliver’s Travels – a device that improved one’s knowledge and helped in one’s contribution to mechanical operations that made even unskilled workers look skilled. Our first introduction to the word robot was found in a 1921 sci-fi play by Karel Čapek who defined this new term as a factory-produced, human-like mechanical person. However, the first programmable digital computer, created in the 1940s, gives a less romantic view of artificial intelligence based purely on mathematics. Called the Atanasoff Berry Computer, it featured a few hundred vacuum tubes that allowed the desk-sized electronic machine to calculate with binary digits, perform logic operations and retain the results in separate memory capacitors. In 1952, a computer learned to play checkers. By 1959, the term ‘machine learning’ described – very simply – a computer that could eventually play chess better than the person that developed the program that ran the computer. 

Artificial intelligence remained a mathematical innovation until the mid 60s when the early natural language processing computer program ELIZA used a script to respond to questions and, the more input it received from different users, the better its responses. This gave us our first taste of personalisation. Chatbots began with Rollo Carpenter’s funny conversational Jabberwacky in the 90s and, also during this time, ELIZA morphed into ALICE. Speech became important in the development of computer programs and early neural networks helped these to understand conversation and intonation.

Artificial Intelligence Today

Today, we are surrounded by artificial intelligence that is far from the ideas movies such as Electric Dreams, Artificial Intelligence and The Terminator have put into our heads. When we look at the actual development of computers that learn from human input, the limitations are more obvious. Mathematics and conversation, logic and gameplay, organisation and memorisation – these are the areas in which AI helps governments, businesses and individuals today.

From carpet-cleaning robots that use sensor-fed data to avoid toppling down the stairs to early forms of facial recognition software which enables border and building security, the future is exciting but not quite as close to the imagination of the screenplay writers as we might hope (or fear). Our accessible AI turns homes into smart-homes, gives the (usually) intelligent content to Siri’s voice, maps our movements so we can play virtual games or travel without needing to look at a map, very gradually replaces our car driving skills, and allows us to communicate with new potential clients by way of sophisticated chatbots. AI is so changing the face of our working lives, many countries are talking about the future possibilities of a Universal Basic Income.

AI in UX

Wherever your business has to deal with a repetitive, boring task, AI software can take over. This also applies to your website visitors. Human website users are spoiled for choice and their every whim is consistently catered to. How willing are you to wait for a web page to load? How often do you fail to contact a business because it only opens from 9 to 5? Do you agree to cookies to avoid having to fill in your name, address, telephone number and email every time you return to your favourite site? Human website visitors are highly critical. Not only do they want a beautifully organised area, they want knowledgeable staff 24/7 to answer questions that border from the inane (how many times has someone asked for a product price from your landing page rather than navigate the website?) to the highly specific. In all of these examples, AI can provide a healthy solution that – at present – lacks the intimacy (and trustworthiness) of the human touch.

AI for Analysis

AI also gives us the means to analyse huge data sets that every separate programme collects. Practically all of your quantitative data research can be done via software and most probably is. Trawling through the data with human staff is simply too time consuming and expensive to ignore the countless solutions available on the Internet. These humans are best put to work with qualitative research (for now) that requires a better understanding of human behaviour and personality. This is where AI is currently making its greatest strides and all personalisation algorithms such as our news feeds and ads on social media platforms are the result of early-stage AI qualitative data analysis. The fact that users need to interact with settings or ads to say whether they find them relevant or not show how lacking in true insight these algorithms are – although by collecting user reactions, the algorithm will learn not to advertise meat products to vegans or gas heaters to people not connected to a gas supply. As the term machine learning suggests, this information must be available for a machine to learn. We do not yet have a wise computer Methuselah. Most algorithms – at least most of those available to businesses as SaaS, hosted or on-premise solutions – are still in the primary school stage. The potential of machine learning is immense, we just haven’t got too far from the starting line yet.

One of the more expensive AI analytics solutions is the use of emotional AI which translates – via cameras on user screens that may or may not be switched on – facial expression. A facial reaction often betrays first impressions and oversteps politeness. A visitor may not be prepared to tell you a webpage makes them feel annoyed but the face rarely lies.

To implement AI effectively on your website, you first need traffic. And to get traffic, you need to tread the rather slow but well-researched road of search engine optimization, quality content, backlinks and guest posts, ads and social media marketing. Quicker methods that only pay off when used in tandem with the aforementioned standard strategies include the use of influencers and opting to pay for additional traffic. Buy website traffic from reputable providers that offer real human visitors or, alternatively, boost your website’s popularity ratings for the Google ranking algorithm and buy traffic Bot software that you can effectively adjust to navigate through your site. To increase the notoriously low conversion rates linked to the human forms of redirected traffic, there’s another option. Buy targeted website traffic that sends large volumes of niche visitors to your chosen landing page (personalised to this particular market) – what you lose in conversion rates, you gain in bigger data and insights that can drive your AI-based UX development.

AI for Design

AI is not yet ready to outcompete the creative human mind. Graphic designers – in fact any creative designer – can breathe a sigh of relief as this career is a safe bet. This was shown on a grand scale in 2017 when Ferrero and Ogilvy Italy set up the Nutella Unica project; they implemented AI to come up with 7 million different labels for Nutella jars with the only red thread being the Nutella spread logo. The often gaudy AI designs showed how little creativity sits within the silicon brain, even though this top marketing strategy had individual jars flying from the shelves as artistic chocolate addicts snapped them up.

What designers can do is integrate AI into their design systems – use programmes to check text alignments with pinpoint accuracy, correctly size content blocks within the design grid, input the visitor journey into a format that the AI solution can filter to make personalised journeys easy and quick. Any well-developed AI design tool will accept scanned sketches and align them with a company brand, for example. This is something that takes time to do without computerised support and pushes up costs.

Add to this AI design features such as multilingual formats of logos and product brochures, the ability to minimalism visitor effort through AI-driven UI generators and real-time testing during web development, correction of grammatical and spelling errors, and the design deadline can be dramatically reduced. This means savings for website owners and less mundane, boring tasks for designers. It does not – and will not – mean a completely computer-derived website design that offers the same level of engagement and visual pleasure as a human-designed site.

AI for Communication

Chatbots are now found on the majority of active websites that deal with higher visitor numbers. As an additional, 24/7 channel for communication, they save time and expense and – for the most part – fulfill consumer need. Most users know how to word a chatbot query – they understand the importance of keywords and the confusion caused by irrelevant terms. Within the next few years, this understanding will grow. Woe to the chatbot that cannot understand ‘shipping price kg UK’ when the potential customer leaves out the term ‘how much is’. While advanced machine learning will one day make this a problem of the past, too many forums (and derogatory memes shared on social media) complain about the inane answers provided by AI chatbots. One size certainly does not fit all. Furthermore, a chatbot will never replace face-to-face communication.

Time to Incorporate AI

Artificial intelligence covers every part of your online presence; listing recommended UX tools for each of these would be an exhausting line of repetitive information. The majority of AI tool developers are bound to tell you how their solution will miraculously elevate your conversion rates – the truth of the matter is that each piece of code will make a slight difference. Just as you don’t place all of your marketing budget into a single channel, focusing on one AI solution is self-limiting. Finding the correct complement of the most effective ones, however, is more than daunting.

When selecting solutions, think of your primary goal. In terms of UX, this should be saving visitor effort and time. The more programs that run in the background, the slower the site – so the temptation of trying out every piece of advertised technology is also not advised. When UX data shows you are on track, find solutions that make you accessible to broader audiences in the form of assistive technology – screen readers, acceptance of alternative input devices and screen magnification, for example. This could bring you head and shoulders above your competition when it comes to communication and the generic chatbot. And always remember to try and test, try and test, try and test. Because until AI becomes the norm, it is a prototype; what works for one business may not work for yours.

Comments are closed.