top of page
  • Writer's pictureJakob Nielsen

UX Roundup: Top UX Articles of 2023 | AI Sings | AI as Provocateur | Integrate AI in Apps | Package Delivery UX | Reporting UX Research | Midjourney Can Spell | UX-AI Job: Ideogram

Summary: Jakob’s top-10 articles and top-3 videos in 2023 | AI sings about the 10 usability heuristics | The future of work: AI changing from an intern to a provocateur challenging users’ assumptions | Don’t bolt on an AI sidekick to applications; integrate it | Lousy UX for package delivery | Midjourney v.6: better spelling and prompt adherence | UX-AI Job: Ideogram design lead

UX Roundup for December 26, 2023. Great tree from Leonardo: rejoice that days are getting longer (my image doubles for celebrating the joys of fall for the ~25% of my readers in the Southern Hemisphere).

Top-10 Articles of 2023

I didn’t start writing until mid-May of 2023, after leaving my former company in April, so I don’t have a full year of articles to analyze. During the 7.5 months I did write, I published 82 articles, which is still enough for a top 10 list to be meaningful.

These are my articles that gained the most traffic in 2023:

  1. UX Angst of 2023 😨 Forget doom and gloom; UX is transforming, not failing. AI spurs a rebirth, assuring long-term growth and opportunity.

  2. Getting Started with AI for UX 🤖 Use generative-AI tools to support and enhance your UX skills — not to replace them. Start with small UX tasks and start now.

  3. UX Vocabulary Inflation 📚 UX suffers from ever-changing terminology, causing miscommunication and rework. It's crucial to adhere to established names.

  4. Dark Design Patterns Catalog 😈 Description of the leading 12 dark UX patterns, targeted by new Indian regulations to eliminate harmful and misleading user interfaces.

  5. User Experience in Two Infographics 👀 Demystify UX for your team and boss in 3 minutes: Two concise infographics explain the why and how of UX projects.

  6. UX Needs a Sense of Urgency About AI ⏱️ UX must rapidly pivot to AI, mirroring the urgency from the dot-com revolution: Adapt to AI or risk obsolescence. Don’t leave geeks in charge.

  7. AI Is First New UI Paradigm in 60 Years 💻 AI UX focuses on outcomes rather than processes, marking a significant shift from command-based design to intent-based interaction.

  8. Ideation Is Free: AI Exhibits Strong Creativity, But AI-Human Co-Creation Is Better 💡 AI's burgeoning creativity outperforms humans in raw idea generation, while human-AI partnerships excel in refining and realizing these concepts.

  9. Jakob’s Law of the Internet User Experience 🌐 Users' widespread web experiences form a baseline expectation, making familiar website designs more usable and effective.

  10. The 10 Usability Heuristics Reimagined 🚀 Revisiting the roots of UX: Jakob Nielsen’s foundational guidelines from 1994 continue to influence contemporary design decisions. See also Jakob Nielsen’s 10 Usability Heuristics as Haikus 俳句

Bonus link: Generative AI Enhances Old Users’ Intellectual Performance Through Wise Winnowing 🧓Enhance senior workers' creativity with AI-produced variety, then distill choices through their wisdom for optimal outcomes. (This didn’t get many clicks. Nobody cares about old people. Read it anyway, because extending the careers of older knowledge workers is crucial for the economy in advanced countries with declining birth rates. We must use the people we have since we’re not getting enough new ones.)

Top 3 Videos of 2023

I didn’t produce many videos in 2023, focusing on my renewed creativity for writing. I featured in a few videos:

  1. The Long View of UX I had an exciting discussion with Felix Lee (co-founder of ADPList, which is the main way to get UX mentorship). The video outlines my shift from management to a creative scientist and writer. We discussed AI's cognitive task potential, drawing tech revolution parallels. The video also tracks Jakob's career through tech changes, emphasizes the importance of the 10 usability heuristics, and considers their modern relevance. Additionally, we explored AI's role in augmenting human creativity, ethical AI use, global UX demand, industry challenges, and the growth in UX and AI integration.

  2. Write Like a Machine, Think Like a Human (Writers in Tech podcast) I believe AI will shift the role of writers from just generating text to being editors and evaluators of text. Future writers will spend more time providing creative direction to AI tools, reviewing outputs, revising based on brand needs, and selecting the best options. The massive productivity gains of AI writing tools mean generating multiple text samples becomes almost free. This facilitates experimentation and creativity. Integrating AI makes individual writers far more capable, effective, and valuable. By mastering AI as a tool, writers can reach new heights of creativity, productivity, and impact.

  3. Advice for Young UXers Working Solo (as the only UX Pro in a company) I recommend being part of a professional team to learn how to do UX right. Sarah Gibbons interviewed me on the occasion of my retirement from my former company. (Sarah is the most talented UX professional I have met in my 41-year career, so she asked exactly the right question to leverage my experience for her audience.)

AI Sings Heuristically

I had the Suno AI music service create two short songs about my 10 usability heuristics.

Here are the lyrics, but do click the links on top of each song to hear the performance set to music. These may not be great songs, but it’s still early days for music AI, and it’s a promising start.

Which of the two songs do you prefer? “Rockin’ Usability” or “The Usability Jive?” Let me know in the comments.


Well, let me tell you 'bout a little thing called usability Jakob Nielsen came around and set our minds free (ooooh-yeah) He laid down the heuristics, ten simple rules to follow To design user interfaces that'll make 'em holler (holler!)


Usability, ooh you gotta make it easy for me (easy for me) Learnability, let it flow right naturally (naturally) Satisfaction, make 'em feel the joy in every click (every click) Efficiency, design it smooth and slick (smooth and slick) (ooh-yeah)


Well, listen up, cats and kittens, gonna tell you 'bout a groove It's the usability jive, gonna make your UI move Jakob Nielsen's the name, with guidelines that are cool He's got ten heuristics, gonna play 'em by the rule


Heuristic one, visibility of system status (ooh-yeah) Don't make it confusing, make it oh-so-precise Heuristic two, match between system and the real world Gotta make it relatable so users feel nice (feeeeel nice)

Unfortunately, one verse and a chorus are as much as they give you for free. For now, I’m not paying to get the music for the remaining 8 heuristics.

AI can now generate songs about any topic you wish. Good for parties, if nothing else. (Dall-E)

The Future of Work: AI as Provocateur

Microsoft has released a great year-end report summarizing what we know about AI use in the workplace and what they expect to happen next. (Warning: 41 slides in PDF format.)

Slides 6-14 are a very useful summary of much of the research that was published this year. I have reviewed and analyzed this research myself in many articles on UX Tigers, but Microsoft’s summary is certainly shorter and faster to read if you need to catch up. As a summary of the summary:

There are many more exciting nuggets in this report: some are based on internal, unpublished research at Microsoft, and some are (informed) speculation by the large group of smart people who wrote the report. I highly recommend downloading and reading the whole thing. It will take almost an hour to read the full document (12,589 words), but since it’s structured as a highly scannable slide show, you can skip to the parts you’re interested in. Here are two highlights, for people who can’t be bothered to download PDFs (I’m with you):

This year, the most common metaphor for AI is the eager but inexperienced intern or assistant, who performs grunt work under supervision. An alternative metaphor advocated in the report is AI as provocateurs in knowledge work, a shift that enhances critical thinking. As AI tackles more creative tasks, it necessitates human expertise in critically integrating its outputs. These AI provocateurs do more than just highlight errors; they challenge users’ established assumptions, prompt evaluations, and present counterarguments. However, the design of such provocative AI must balance offering constructive criticism without overwhelming users. Furthermore, interactive technologies that stimulate discussions and engage users play a crucial role in fostering critical thinking skills. I would add that this perspective seems like a great way to employ AI in education.

Traditional metaphor: AI as an eager intern who can do simple tasks under supervision. (Dall-E)

Alternate metaphor: AI as a provocateur to challenge and maybe change knowledge workers’ thinking. (DallE)

Internal data from Microsoft shows that generative AI (Bing Chat) is used differently than traditional search engines (Bing Search). The data shows that a significant majority of Bing Chat conversations, about 69%, are related to professional domains, which is almost twice the 39% observed in Bing Search sessions. Additionally, Bing Chat sessions exhibit a higher level of complexity. Specifically, 36% of AI conversations involve tasks of higher complexity, such as application, analysis, or creation, which is almost 3x the 13% of high-complexity tasks in search sessions, indicating a shift in how people use these platforms for more sophisticated and professional purposes.

Don’t Bolt an AI Sidekick onto Your Application; Integrate AI

The two most talented UX professionals I know, Sarah Gibbons and Kate Moran, appeared together in a webcast about AI&UX hosted by the online UX Design Masters’ Program of Atlantic Technological University in Ireland. (I don’t know anything about this degree offering other than they must have a finger on the pulse to host such excellent external speakers. Given the online nature of the degree, they target an international audience, which I guess is why they have international speakers.)

The full video recording of the webcast is available (44 minutes, though I recommend starting at the 4-minute mark where the real discussion begins.)

Some of the key points discussed were:

🎨 Integration of AI in UX with a focus on context and curation

🚀 AI enhancing UX efficiency, particularly in ideation and project finalization

⚙️ Future of AI in UX: Integration into existing tools and processes

🌿 Impact of AI on UX roles and the importance of maintaining creative and critical skills

🎈 Use of AI in personal projects, such as party planning, to understand its capabilities

🎯 AI's role in accelerating UX research and its limitations in data analysis

🛡️ Ethical concerns in AI, particularly in terms of the privacy of user data

🤔 The importance of critical thinking and experience in discerning AI outputs in UX

📚 Encouraging UX professionals to adapt and integrate AI into their workflow

One particularly interesting issue was a warning against simply bolting on a piece of AI on the side of an existing application. There’s a temptation to add chat, even though it’s often a poor way of supporting users’ tasks. As Moran pointed out, often, company leadership simply tells their teams to “put AI somewhere in the product.” Gibbons characterized this as “abandoning the point of our job, which is identifying what the actual problem is, instead of jumping to the solution.”

Bolting a chat window onto an existing product will not create a usable and seamless user experience (Dall-E)

The speakers emphasized that the better solution is to seamlessly integrate AI within each domain-specific product with features optimized to making actual user tasks more efficient.

Gibbons had found AI to be particularly helpful at the divergent beginning of a design project and at the end of convergence. In contrast, the middle part depends more on traditional human work. Both speakers emphasized the value of leaning into the randomness of AI for ideation: don’t be afraid of bad AI results because you just discard those parts if you had asked it for 20 alternatives.

Lean into the randomness of AI for creative ideation. (Dall-E)

Package Delivery UX

The last step of the e-commerce user experience is getting your package delivered. Until you lay hands on that box, it’s all hypothetical.

Peter Ramsey reported on the experience of shipping and receiving packages with 6 different delivery services in the UK. Most had a terrible design for ordering a delivery, which is a problem for small companies without a vast shipping operation with its own software for fulfillment management.

The experience for the recipient is also bad, primarily due to a lack of communication. Only one of the services provided real-time communication about the status of the delivery and the expected delivery time. This service (DPD Local) also scores the highest in customer satisfaction, even though Ramsey cites data that its actual delivery performance is no better than what’s provided by competitors with a third of DPD’s satisfaction score. While this experiment is too limited to know for sure, it’s likely that the improved digital experience of keeping customers informed substantially impacts satisfaction. Setting accurate expectations is a powerful component of UX.

Package delivery: many services don’t set expectations for the recipient as to when they will get their package. (Midjourney)

Midjourney v.6: Better Spelling and Prompt Adherence

Midjourney has always been one of my favorite image-generation tools because of its high image quality. Just look at that delivery scene above! This despite Midjourney’s abhorrent usability with many blatant UX mistakes. (Many of these design mistakes look to be fixed in the new web interface that’s currently in Alpha.)

One downside has been Midjourney’s complete inability to spell words, which we often want to include in AI-generated artwork. It also had low prompt adherence for any requests that included more than one object. For example, the seemingly very simple request for “a doctor at the bedside of an injured patient with a robot watching from the corner” didn’t work at all, though Dall-E produced a great version for “Artificial Empathy: Is It Still Empathy?

Midjourney version 6.0 is now available, though it’s still experimental and you must enable it through the /settings command to try it out. Much improved! And the production version will presumably be better. Though probably still not what I really want. Let’s look at some examples.

First the hospital scene, where a robot (being the stereotypical visualization of AI) is helping a doctor with poor bedside manners exhibit more empathy toward an injured patient. After some prompt modifications, I succeeded in getting Midjourney to give me this image with all three characters, but it doesn’t tell the story nearly as well as the image I got from Dall-E (in two attempts) and ran in the article.

My favorite slogan about learning how to use AI in UX is “Start Small, Start Now.” This is a successful image from Midjourney, even if the text is a little too small in my opinion. Much more interesting than the attempts at the same slogan I got when testing Ideogram version 0.2.

Two versions of the classic UX slogan, Keep It Simple. For the top version, I asked Midjourney for “elegant typography,” whereas the bottom version was generated by prompting for “clean simple typography.” The overdesigned (though cute) top version certainly doesn’t represent the message very well, whereas the bottom version does. (Except for the unexplained extra occurrence of the slogan’s initial word.) While simple, this second image doesn’t give us anything extra to justify being an illustration, as opposed to just writing out the three words in the body text.

Finally, I attempted Sarah Gibbon’s challenge of illustrating the UX slogan, “It Depends.” Midjourney did give me a workable version of a bi-directional road sign featuring the slogan, but I don’t like it as much as the design Sarah got from Ideogram. (And even less than her human-designed version. She’s a typography master, whereas AI is still at the apprentice stage in typography.)

After this small experiment, the score is a 1-1 tie between Midjourney v.6 and Ideogram v.0.2 for generating images that include text.

UX-AI Job: Ideogram Design Lead

As just noted, Ideogram is still in its very early stages, so I applaud them for hiring UX professionals. (Though a few months ago would have been better.) They have an opening for a Design Lead. The company is based in Toronto, Canada.


Top Past Articles
bottom of page