In 2025, AI video models made significant advances, especially in avatar expressiveness, allowing me to create far superior videos compared to 2024. Despite progress, 2026 will require further improvements for individual creators to fully harness the technology’s potential. Below are my 10 most popular videos of 2025 based on audience clicks and viewing times. I initially believed 2024 marked the year of AI video and 2025 would be the year of AI agents. While AI agents grew notably in 2025, they remain immature for widespread use, though enterprise applications are emerging. AI video saw modest progress in 2024, exemplified by my music video “Top 10 UX Articles of 2024. ” However, 2025 brought major leaps in capabilities, making earlier videos appear primitive in retrospect—for example, “Old Workers Stay Creative With AI” (December 2025). To illustrate 2025’s progress, I created a highlights reel featuring clips from my best music videos released throughout the year, presented chronologically to showcase improvements. Regarding specific AI video components, improvements varied: - Speech synthesis exhibited modest gains since it was already strong by late 2024. The biggest step was models like ElevenLabs v3, which use language understanding to add emotional nuance to speech—for instance, in my explainer “Slow AI: User Control for Long Tasks Explained in 5 Minutes. ” - Songs and music also improved modestly. Notably, I produced my first operatic aria that sounds reasonably good (“Direct Manipulation”), overcoming prior failures that resembled poor Broadway musicals. - Avatar animation advanced considerably; models like HeyGen Avatar IV deliver high-fidelity, particularly in close-up talking head views, though full-body avatars still suffer quality losses. The music video “Creation by Discovery: Navigating Latent Design Space” shows much better singing performance in closeups. - Dance and movement animation improved but remain insufficient. For example, the K-pop style dance sequences in “Creation by Discovery” have good individual moves but lack cohesive choreography and synchronization with music. Movements of singers and musicians also fail to coordinate properly with audio, limiting realism. - Native audio-video generation saw a huge leap with Veo 3 and 3. 1, generating synchronized 8-second clips ideal for B-roll but too short for full videos. As an experiment, I stitched clips for an AI rendition of Shakespeare’s “Pericles, Prince of Tyre, ” which conveyed plot basics but falls short of enjoyable storytelling. Overall, all AI video areas progressed in 2025. Looking ahead to 2026, despite some claims, traditional Hollywood is not obsolete—fully AI-generated mainstream film or TV remains years away, perhaps by 2028. Studios like Amazon effectively use AI for scenes and effects, but legacy workflows persist. I anticipate major studios will either transition to AI or struggle by 2028. Expect further improvements across the AI video pipeline in 2026, especially in native audio generation, potentially extending fully synchronized clips to 30 seconds. Character consistency between clips is possible visually by referencing still images (e. g. , in “Aphrodite Explains Usability” using Veo 3. 1), but vocal consistency remains lacking, making sustained narrative sequences less believable. AI voice consistency is likely achievable in 2026 to enhance storytelling flow. Music generation is important for videos and music videos alike. One-shot songs created with Suno 5 are already enjoyable—personally, I prefer listening to my Suno-made songs over chart hits, partly because AI enables personalized creative freedom not limited by corporate tastes. However, existing music models offer only limited, technical editing features like equalizer adjustments. We need semantic editing allowing higher-level expressions, such as requesting softer drums or emphasizing specific instruments to evoke moods (e. g. , spooky versus romantic), enabling intuitive navigation of musical latent spaces. This need for meaningful semantic control applies to all media elements involved in video creation. In summary, 2025 marked strong advances in AI video but remains insufficient for ambitious projects.

Still, the future is promising. For determining my top videos of 2025, I analyzed several metrics: - Clickthrough rate measures how many users click a video’s thumbnail, influenced more by thumbnail design than video quality. I avoid clichéd YouTube tropes (like exaggerated astonished expressions) despite their effectiveness, resulting in moderate click rates. - View count tracks users who watch beyond just clicking. I combined views across platforms (YouTube, LinkedIn, Instagram, X), adjusting for video age on YouTube since older videos have longer to accumulate views. - View duration is a strong indicator of quality. YouTube provides average view time and percentage watched, but both have limitations—average view time favors longer videos; short videos can’t reach high averages even if fully watched. I prefer analyzing drop-off rates; for example, measuring viewers remaining after 30 seconds shows engagement normalized across video lengths. Retention curves for two versions of my “Direct Manipulation” song showed viewers favored the rock version over the opera. Moving forward, 2026 may unlock AI video’s ability to empower independent creators more fully and could usher in transformative shifts from linear storytelling toward immersive worldbuilding, where users become active participants. My Top 10 Videos of 2025 are: 1. No More User Interface Explores AI’s evolution replacing traditional UI with deeper experience orchestration as AI agents handle more tasks. 2. Service as Software (also a music video) Discusses AI intelligence becoming freely available, transforming services into instant software and enabling massive skill scalability. 3. Making a Usability Action Figure A playful video showing a 3D animated action figure created via AI, illustrating fun applications of technology. 4. Vibe Coding and Vibe Design (also music video) Demonstrates AI’s impact on software and UX design by enabling natural language intent use, accelerating prototyping while emphasizing ongoing human expertise. 5. UX in 2025: Jakob Nielsen’s 6 Big Themes (also music video) Presents forecasts for major UX trends in 2025 by Jakob Nielsen. 6. UI vs. UX: Jakob Nielsen Explains the Difference Clarifies the distinction: UI comprises tangible elements users interact with, while UX is the overall user satisfaction shaped by UI and other factors; AI is expected to automate most UI design allowing humans to focus on UX strategy. 7. Recognition Rather than Recall (Jakob Nielsen’s Usability Heuristic 6) Highlights the importance of minimizing users’ memory load by making information easily recognizable rather than requiring recall. 8. Pivot Your UX Career for the AI Age Advises UX professionals to develop uniquely human skills (agency, judgment, persuasion) as AI automates legacy tasks, urging adaptation rather than resistance. 9. Error Prevention Explained by Vikings (Jakob Nielsen’s Usability Heuristic 5) Emphasizes designing to prevent errors using constraints, validation, and risk-scaled friction to build user confidence and reduce support costs. 10. AI Going Mainstream: Crossing the “Chasm” to Early Majority Users (also explainer) Analyzes AI’s uneven adoption shift from early adopters to mainstream users globally. Bonus: AI Helps Old Users Stay Creative (available as music and avatar explainer) Discusses how AI compensates for age-related fluid intelligence decline in older creative professionals, extending their productive creative years despite limited popularity among younger audiences. Main takeaway: UX is evolving rapidly alongside AI innovations—are you prepared? About the Author: Jakob Nielsen, Ph. D. , is a renowned UX pioneer with over four decades of experience and founder of UX Tigers. Credited with ushering in discount usability and formulating key usability heuristics and Jakob’s Law, he is widely called “the king of usability. ” He holds 79 US patents, authored eight influential books, and earned prestigious awards including ACM SIGCHI’s Lifetime Achievement and Human Factors Society’s “Titan of Human Factors. ” He has served as Distinguished Engineer at Sun Microsystems and researcher at Bell Communications Research. Subscribe to Jakob’s newsletter for new articles, follow him on LinkedIn, or read a detailed article about his UX career.