Artificial Intelligence and Web Design


Hello, World. I am your web designer…

Recently, I read an article about using artificial intelligence (AI) for the instructional design of courses. Initially, that frightened me. First of all, it might mean less work for instructional designers – which I have both been and run a department working with them. Second, it’s hard for me to imagine AI making decisions on pedagogy better than a designer and faculty member.

Of course, using AI for that kind of design is probably limited (at least at first) to automating some tasks like uploading documents and updating calendars rather than creating lessons. Then again, I know that AI is being used to write articles for online and print publications, so who knows where this might go in the future.

I just read another piece asking “Is Artificial Intelligence the Next Stepping Stone for Web Designers?” and, of course, my concerns are the same – lost jobs and bad design.

Certainly, we are already using AI in websites, particularly in e-commerce applications. But using AI to actually design a website is very different.

Some companies have started to use AI for web design. A user answers some questions to start a design: pick an industry or category (portfolio, restaurant, etc.), enter a business name, add a subtitle/slogan/brand, upload a logo, enter an address, hours of operation, and so on. The AI may offer you a choice of templates and then in a few clicks, the basics of the site are created.

This is an extension of the shift 20 years to template-driven web design. Now, it is based on machine learning techniques with human intervention at the initial stage by providing their desired information and probably again after the site is created to fine-tune.

I do a lot of designs in Squarespace and they are clearly using AI and machine learning to get you started. Do you still need human intervention? Absolutely. Does the human need to be a “designer”?  Clearly, the goal is to allow anyone to do a good job of creating a website without a designer.

In my own work, I still find many people need someone with experience and training to create the site, but they can oftentimes maintain it on their own if the updates are simple. I have also had clients who with just a few clicks have completely wrecked their websites. And there is no Cntrl-Z or Undo button to put it back together again.

AI will change – dare I say revolutionize – many industries and design is certainly on the list. When AI can make the process more efficient, I am all for it, but I stu=ill like the human factor in any design project.

Can We Measure Social Media Sentiment?


Social media sentiment is the perceived positive or negative mood being portrayed in a social media post or engagement.

If you could track sentiment accurately, it would help you understand the person’s feelings behind the post. As a marketer, this would be very useful, but would also be useful for individuals.

Certainly, there is sentiment behind every post. But can it be measured with any certainty?

There are scholarly articles about social media sentiment and a good number of companies that are working on trying to measure sentiment, which you might also see described as sentiment analysis or opinion mining. By any name, this is the analysis of the feelings (i.e. attitudes, emotions and opinions) behind the words.

Most of the tools use natural language processing. Natural language processing (NLP) is a field of computer science, artificial intelligence, and computational linguistics concerned with the interactions between computers and human (natural) languages.  It allows you to talk to your phone or some device in your home or car. These devices can understand your words (usually), but can they read your emotions? When you ask for directions are you angry, tired, frustrated or in a wonderful mood?  When someone responds to an offer posted in social media by a company by saying “This is crazy!” is that a good kind of crazy or an insanity crazy?

In social media, this is analysis that goes beyond Likes, Shares or Comments. Did people respond to the original post in a positive, negative, sarcastic, humorous or biased way? A human reader may be able to discern that, but can a tool do the same thing for hundreds, thousands or millions of posts?

The  complexity of emotional responses makes this analysis difficult. You have heard a lot lately about sites like Twitter and Facebook being told that they need to better monitor hate speech in their networks. How do you do that? Rely on users to report it? Have other humans monitor it? That won’t work when every second, on average, around 6,000 tweets are tweeted on Twitter, which corresponds to over 350,000 tweets per minute or 500 million tweets per day.

You need technology, but technology is famously not very good at reading human emotions.

There are some simple tools that some of you might already use for analysis.  Hootsuite Insights and Facebook Insights and SocialMention are some of the easy and more common free(mium) tools for analysis, but they are lacking in the analysis of sentiment. Many businesses and individuals use Google Alerts as a simple way to monitor their name, brand, and to track “content marketing” with the result being emailed as they occur, daily or weekly.

We are still a good ways off from a time when some combination of NLP and AI can read the sentiments of social media posts accurately, but the desire and need for it only grows more critical as networks grow.