AI: The Intelligent Web

Grant Russell, Developer

AI: The Intelligent Web

Artificial intelligence, chat bots and the internet of things have been hyped for a few years, but at SXSW 2017, it was evident that these technologies are becoming an essential part of the modern web. Industries ranging from health to news and entertainment are using smarter technology to connect with their customers on a more personal and relatable level.

Conversations

A key theme at SXSW was the recognition that AI is going to change the way we interact with online services. Almost every speaker and tech professional were talking about it and their enthusiasm radiated throughout the event.

I learned a new abbreviation that week: CUI (Pronounced coo-wee) which means Conversational User Interface. These types of interfaces involve the end user communicating with web services using natural language in a question and answer fashion.

Conversational Search

A design lead at Google ran an inspirational session where he demonstrated the future of conversational platforms. He played out a scenario of arranging a trip to SXSW using voice. The system would ask several questions like “Do you have an airline preference?” and “How long will you be staying?”. The system asked around ten simple questions and from the information that was gathered, it was then able to book everything from flights, to hotels and event tickets; all without the user filling in a single form. This future is still a few years away from being mainstream, but this session conveyed a very clear message that search as we know it will change dramatically. Currently, using a search engine returns hundreds and thousands of results, whereas future trends are going to be that the search engine asks questions back to get a more contextual understanding of the original question. From this information, irrelevant material can be removed from the search results and only the relevant results are then displayed.

Chat Bots

I was excited by the future prospect of CUI’s, but I wanted to understand how they (specifically chat bots) can be used today and what value they will add to a service. I went into each session with a very critical mindset, however, every speaker managed to strongly convince me that chat bots are a necessary progression for the web.

The key thing that chat bots strive to achieve is to allow a user to gain frictionless and instant access to information whilst maintaining a personal and friendly brand relationship. An example that was mentioned in one presentation is that you are currently expected to phone a company to get an immediate service. This experience is frustrating when you must wait on hold for twenty minutes and navigate through complicated menus. Furthermore, a phone call is very personal and people feel slightly uncomfortable with this; especially emerging younger audiences, who would often only text their friends and rarely call. Chat systems allow companies to maintain an instant and personal service, whilst removing frustration and the intimacy that a phone call creates.

Of course, chat windows with humans have existed for years, but what is different now is that the chat can be automated easily for its entirety, or partially with a handoff to a human operator when required.

The key things that I brought back with me from SXSW are that there are three types of chat bot:

  • Basic FAQ Bots
    This type of bot simply returns a relevant answer for any given question.
  • Advanced Search Bots
    This type of bot is a little smarter than the FAQ bot, in that it has the ability to ask questions back to narrow down results.
  • User Journey Bots
    This type of bot replaces or works alongside a traditional GUI (Graphical user interface) and guides the user through the process.

Probably the most exciting thing about chat bots is that they can be integrated into a website on the web or added as a contact on many popular chat platforms. One key point is that downloading traditional native mobile apps requires a certain amount of investment for users. They must first open an app store, log in, then download the app using valuable mobile data and storage. Once all that is done they can open the app and register etc… The experience of chat bots is different; It is as simple as adding a friend on a messaging platform.

Personalisation

Aside from chat bots and conversational UI’s at SXSW, there was a strong correlation between sessions that some level artificial intelligence and machine learning is required in all modern applications. Whilst there is nothing new about consumers expecting a personalised experience, there is a current drive to move to ultra-personalisation. The first step to achieving this personalised experience is to capture as much relevant data as possible in a way that is lawful and ethical.

At one session, I learned how Yummly (a recipe mobile app) have a detailed sign-up process, however, they use a smart mix of language and design to convey the importance of this stage. Typically, users don’t like long sign-up forms, however, they don’t mind spending time setting up their profile if they understand that they will get value back in the form of a better user experience. Yummly then uses this data to personalise results by suggesting content which the algorithm believes they will be interested in.

An important consideration when building personalisation systems, which is often missed; is not to create a “personalisation bubble” where users only see recommended content. Over time this bubble can narrow down smaller and be detrimental to user interest and retention. Facebook has had this problem for a while in that it promotes content that you are interested in and subsequently re-enforces your own worldview, regardless of the diversity that actually exists. Yummly and Spotify both get around this issue by implementing an “Explore vs Exploit” system. In this system, you want to exploit a user’s interests by suggesting mostly recommended content, but ensure that a small percentage of content is strategically random or risky. If users click and consume the random content, then this must be fed back into the machine learning algorithm to adjust that users’ future recommendations.

“A key thing to remember is that you are not only presenting content that the user likes today. You need to present random content to work out what the user wants tomorrow”

Smart CMS’s

I was pleased to see companies at SXSW are integrating AI into their CMS’s to automate some or all the process of creating and displaying content. One company explained how their entire front page of their website was managed by AI. The system considers the user profile information, social media (personal and collective) activity, date and time and promotes content dynamically to maximise the relevance of its site. Other features of this CMS included automatic hierarchy organisation and content moderation. This kind of CMS is going to become a much more common occurrence on the web and can be achieved with relatively simple technologies. There was also a lot of discussion around future SEO, whereby search engines are now looking for the micro-relevance of a site. IE: Is your site relevant given the aforementioned data factors. To enable this, companies are looking at how they structure content and display it in a contextual and semantic way.

In Summary

AI has been rising for a long time, but the tools and technologies are now more accessible than ever. There are amazing new opportunities to engage with end users more personally and efficiently with chat bots and voice conversational interfaces. Personalisation is becoming, even more, the norm across even more sites and services. Suggested content is just as important as recommended content. CMS’s are becoming more automated and will consider a lot of different factors when promoting content.

Returning from SXSW has left me buzzing with excitement to begin working on next generation applications. It also raised questions many questions; some of which we have already started working on. Stay tuned.

We live in a world where almost everything is somehow connected with a brand or presented to us via a brand or a branded platform. And much of the content we consume is enabled or enhanced by technology. This is the world in 2017: brand and technology are the two most important factors in the way we experience and make sense of the things around us. 

TheTin is a Brand and Technology agency. Visit us at www.thetin.net

Comments are closed