Summary: Saving fuel by AI-optimized airline routing | Apple violates 40 years of GUI standards by making checkboxes round in the Vision Pro headset | Recommended reading about storytelling and non-UX topics for UXers | More jobs & higher salaries for people with AI expertise | Perplexity now a startup “unicorn” worth $1B
UX Roundup for March 15, 2024. (Leonardo)
Saving Fuel by AI-Optimized Airline Routing
Andreessen Horowitz (better known as A16Z) has released a report titled “The American Dynamism 50: AI Edition.” As the name implies, they profile 50 startups in the United States with products based on AI. Many interesting small stories, including:
Air Space Intelligence uses AI to optimize airline routing based on a combination of real-time and historical data. Using the platform, Alaska Airlines has shaved 5.3 minutes off the average flight. Maybe this doesn’t sound like much, but it sums to millions of pounds of airline fuel saved per year (with an equivalent drop in emissions).
AMP Robotics uses computer vision to sort recyclable waste 4 times faster than humans.
Farm-ng makes an all-electric micro-tractor that can be modified for an array of cropping and harvesting jobs, including mowing, seeding, compost spreading, cultivation, plant breeding, and more.
Gecko Robotics makes small inspection robots that check equipment for signs of degradation and gather detailed data. The technology is used to evaluate critical infrastructure for energy providers, as well as to track maintenance cycles and identify necessary repairs for U.S. Navy ships.
Common for all 50 case studies is that they are highly domain-specific and often quite nerdy. This is in contrast to the large language models (e.g., ChatGPT) and image-generation tools (e.g., Midjourney) that fire the imagination and, admittedly, have been the focus of my own work. It’s easier to analyze new technologies that target knowledge work and office professionals.
This is similar to the way word processing was historically the “lab rat” of early human-computer interaction research. It was easier to study the usability of document creation (then or now) than to study the work performance of, say, telephone company specialists provisioning central offices with switching equipment. (Though I did a little of the latter when I worked at Bell Communications Research.)
To be honest, I won’t be visiting a bunch of farms for field studies of how the farmers use the micro-tractors. However, this work needs to be done to improve the UX of these specialized products.
Our next frontier is the user experience of highly specialized AI tools, such as tiny inspection robots for electricity infrastructure. (Midjourney)
Checkboxes Are Square; Radio Buttons Are Round
For more than 40 years, one convention has held firm for graphical user interfaces (GUI): checkboxes are square, and radio boxes are round. (Nikita Prokopov has a great historical recap of radio buttons vs. checkboxes, with screenshots of these GUI widgets stretching back to Turbo Pascal from 1992.)
A brief recap for readers without a firm grounding in GUI design:
Checkboxes are used for lists of selection options where the user can select as many or as few as desired, from zero up to the full list. Checkboxes are square, and when selected, the square is populated with a checkmark (or, sometimes, an X).
Radio buttons are used for lists of selection options where one and only one item can be selected at any given time. Radio buttons are circles, and the one selected option is indicated by a large dot inside its circle.
This distinction between having to select exactly one and being able to select any number is useful, and it’s also useful if users can tell at a glance which of the two applies to any given selection list.
The GUI standard has always been the same across platforms, though there has always been a few deviant websites that refused to play along.
Now, to its eternal disgrace, Apple has decided to make the checkboxes round in the user interface for its new Vision Pro headset. This will do no good, but a lot of harm, because deviations from the standard undermine users’ ability to form mental models and predict how the UI will work.
Squares are for checkboxes, and circles are for radio buttons. Not the other way around. Sticking to this standard empowers users to be more confident in using your design. (Midjourney)
Good Books About Storytelling and Non-UX
Nancy Duarte is the world’s leading expert in creating business presentations. She has long emphasized the benefits of relying on storytelling when crafting compelling and persuasive presentations. Duarte has now published her list of 11 favorite books from which to learn the storytelling craft. Her top recommendation is The Writer’s Journey by Christopher Vogler, but you should read them all if you’re serious about communicating your UX work to teams and stakeholders.
Good old-fashioned books create a more immersive reading experience than online media. Perfect when you really want to learn something. (Leonardo)
Lawton Pybus gave us a completely different set of reading recommendations in his recent newsletter, “The Quarter-Inch Hole.” (A highly recommended newsletter about UX methods, by the way. It’s named after the saying that users supposedly don’t want a drill; they want a quarter-inch hole in the wall.)
Do you want a drill, or do you want a hole in the wall? (Midjourney)
Pybus crowdsourced a list of 25 recommended books from his readers (including yours truly). The twist? These were not UX books, of which we have many reading lists already. These were non-UX books that UX professionals recommended to their colleagues as helpful. The top recommendation was Thinking, Fast and Slow, by Daniel Kahneman, which explains the cognitive biases humans evolved to make fast decisions when pursued by a sabertoothed tiger. And which we still use in business settings today, much as the tiger is absent.
If this guy couldn’t make fast decisions that worked, he would not be among our ancestors. (Midjourney)
More Jobs & Higher Salaries for People with AI Expertise
This will not come as a big surprise, but it’s nice to see some quantification estimates. The Wall Street Journal ran a story last week titled, AI Talent Is in Demand as Other Tech Job Listings Decline. (Subscription required.)
December 2022 was more or less the end of the Covid-induced hiring bubble. Since then, new job listings specifying AI experience are up by 42%, whereas overall IT job listings are down 31%. Thus, relatively speaking, AI jobs are more than twice as hot as non-AI tech jobs. (142% is more than two times 69%.)
Eyeballing a chart in the article, tech jobs with an AI component offer about 15-20% higher salaries than jobs in the same field without an AI component.
No kidding, Sherlock!
AI skills are currently a veritable fountain of gold in the job market. I told you so. (Leonardo)
The underlying difference is still worth a little analysis, much as the conclusion that AI experience is beneficial to your career is so elementary that Dr. Watson could have made it.
All tech jobs are down from the peak (as are UX jobs). In general, AI improves knowledge-worker productivity by about 40% (which is also my estimate for UX jobs). Still, programmers are so far the profession to benefit the most from AI, being able to more than double their productivity while coding. (Being a developer also involves many non-coding tasks, so their full-day productivity will increase less.)
This is why companies are hiring less IT staff right now. They prefer not to lay off too many old-timers with experience in the company’s codebase, and thus are reluctant to bring on new programmers.
We are currently in the stage of the AI revolution where companies are cashing in the productivity gains by being able to do the same work with less staff. We are not yet at the next stage where companies stop doing the old work and start reorganizing the way they operate to profit from the new opportunities. When programming is at half price, and UX is much cheaper as well, companies can create more sophisticated products with advanced software and dramatically enhanced usability. Once we reach this stage, the real expansion will begin with substantial job creation for both geeks and UX’ers. That’s also when we’ll get a real jump up for the world economy to pull us out of the current slump.
Perplexity = $1B
One more piece of news from last week's Wall Street Journal. The AI-driven answer engine Perplexity now has a stock valuation of $1 billion, which is the threshold for so-called startup unicorns. More strikingly, this is double the company’s valuation from only two months ago.
The article also reports that user traffic increased by 8.6% in February, corresponding to an annualized growth rate of 182%. The company now has 50 M MAU.
Perplexity is now a startup unicorn (i.e., worth $1B). (Dall-E)
To give you an example of the benefits of using an AI-driven answer engine instead of a legacy search engine like Google, here’s what Perplexity gave me when I asked it to define MAU. My main usability suggestion would be to highlight the actual definition of MAU, which I have done for it in this copy. Perplexity itself only boldfaces the subheadings.
Monthly Active Users (MAU) is a key performance indicator (KPI) employed across various online platforms, including social networking sites, to quantify the number of unique users who interact with a website or platform within a given month. This metric serves as a fundamental gauge for assessing the health, growth, and popularity of online services. It is instrumental in evaluating the effectiveness of marketing campaigns, retaining existing customers, and attracting new ones. Despite its widespread use, the calculation and interpretation of MAU are not standardized across the industry, leading to potential discrepancies in its application and understanding.
Definition and Calculation
MAU tracks the number of distinct users who visit a website or engage with a platform over a specific period, typically a month. This measurement is crucial for determining the performance and growth of online sites. However, the lack of uniform standards for defining what constitutes a "user" and what qualifies as "active" introduces variability in how MAU is calculated across different companies. For instance, some businesses may count a user as anyone who has accessed their site, while others may require a login and password creation, or even specific interactions defined by the company, to consider a user as active.
Importance and Limitations
MAU is considered a vital metric for online businesses, particularly for those in social networking, online gaming, and mobile applications. It provides insights into user engagement, growth trends, and the overall health of the product or service. Investors and stakeholders often scrutinize MAU figures as they can influence a company's valuation and stock price.
However, the metric's utility is subject to debate due to its inherent limitations. The absence of industry-wide standards for MAU calculation can lead to unfair comparisons among competitors. Moreover, MAU alone does not account for the depth or quality of user engagement, making it a potentially superficial measure of a platform's success. Critics argue that MAU should be considered alongside other metrics to gain a more comprehensive understanding of user behavior and platform performance.
Practical Application and Criticism
Despite its limitations, MAU remains a widely used metric. Companies like Meta (formerly Facebook) and Twitter (now X Corp) have employed MAU and its variants (e.g., DAU for daily active users) to track user engagement and growth. However, the relevance and accuracy of MAU as a standalone metric have been questioned, prompting some companies to adopt alternative or supplementary measures, such as monetizable daily active usage (mDAU) or "Family" metrics that account for users across multiple platforms.
In conclusion, while MAU is a valuable indicator of user engagement and platform growth, its effectiveness is contingent upon the context in which it is used and the additional metrics it is paired with. The lack of standardization and the metric's inability to measure user engagement depth highlight the need for a nuanced approach to interpreting MAU figures.