What Defines ‘Good’ Now?

Explorer making sense of the connections — image via Runway ML

This article is part of ‘Cultivating Creative Intelligence: An Article Series for the AI-Driven Future.’ In this series, we explore how AI is redefining creativity, design, and leadership, drawing insights from the Multiverse Machine project, the CTRL+ALT+EVOLVE movie, and responsible Agentic AI research. (Catch up on Part 1: https://www.linkedin.com/pulse/creative-exploration-vibing-ai-lee-ackerman-0pcic and Part 2: ). https://www.linkedin.com/pulse/creative-multiplier-scaling-impact-through-true-lee-ackerman-r9zlc

Section 1: The Challenge of Defining ‘Good’ in the AI Era

The rapid rise of AI is fundamentally reshaping industries and practices. As digital workforces scale capabilities and accelerate iteration at unprecedented speeds, a fundamental question emerges that has profound implications for how we build, how we lead, and how we define success: What Defines ‘Good’ Now? We are no longer just optimizing for efficiency; we are redefining the very essence of quality, value, and human-AI collaboration in an increasingly automated world. This shift challenges us to look beyond traditional metrics and truly understand what constitutes excellence in an age of abundant, AI-driven output.

To understand “good” in the AI era, we must first acknowledge that this transition isn’t purely about logic, efficiency, or productivity. As explored in my earlier article, “I ❤ AI! Or Do I?”( https://www.linkedin.com/pulse/i-3-ai-do-lee-ackerman-3bhrc ), we cannot overlook the emotional and human side of AI. Our perceptions are shaped by a dichotomy of hope and fear, influenced by cultural narratives from movies, books, and art. Concepts like remediation highlight a natural tug-of-war between old and new ways of doing things. The shift isn’t just technical; it’s deeply human and cultural.

Section 2: Lessons from the Multiverse Machine Experiment

The Multiverse Machine project became a hands-on investigation into what “good” means when creation happens at speed, with AI as a partner. We focused on true collaboration — bringing together a team of different tools, personas, and capabilities. This team could be available when needed, be extremely productive — but also highly creative and provide unexpected (but yet valuable) contributions.

The Power of Unexpected Outputs: The Post-Credit Scene

A favourite example of an “unexpected contribution” is the development of a post-credit scene foreshadowing an intriguing sequel for CTRL+ALT+EVOLVE. This sequence emerged from outputs that initially seemed like ‘failures’ or misinterpretations of creative intent.

For instance, prompting for forest animals encountering holograms yielded an image of a beaver in a lab, surrounded by computers and a hologram-a fascinating, if unrequested, departure. Similarly, a request for a rabbit in a forest storm produced a surprisingly menacing figure, and branches at night during a storm became unsettling gates of claw-like branches.

These visuals, initially set aside as merely ‘cool’ or ‘interesting,’ later leapt to mind as a mentor suggested adding a post-credits sequence to the movie. These unexpected visuals were the creative spark for a dark, conflict-laden story hinting at a villain and the potential fallout from adopting technology without guardrails. This demonstrated how AI’s diverse and sometimes ‘unwanted’ outputs can, with a fresh lens and a willingness to re-contextualize, become powerful assets that surprise and add unexpected narrative depth.

Vibe Coding and Rapid Iteration

The version of the app that you might explore today is not where it started. Through vibe coding-an intuitive, rapid prototyping approach that prioritizes emotional resonance and conceptual alignment over strict technical specifications-and the power of AI tools, the initial concept went from a paragraph of an idea to a functional app with uniquely created worlds — complete with aligned visuals and vocabulary — in a couple of days of effort. This rapid realization allowed for immediate testing and, crucially, immediate feedback.

This approach implicitly aligned with principles of customer development and validated learning, allowing us to quickly ascertain if we were building something truly valuable, not just rushing to a solution.

The Evolution of ‘Good’: Feedback and Provocation

In early demos, other creatives exploring the app began questioning the UI (and they still probably could!). But the more interesting feedback emerged around some of the images. One particularly memorable moment came with the pointed question: “What were you thinking with the composition of this image?” I knew they were using that word, “composition,” in a specific way that I initially didn’t fully grasp. As we talked, they kindly shared some background, helping me understand the principles they were referencing.

This feedback sparked a critical “a-ha!” moment. While we discussed the AI generation of the image, the more profound realization, for me, was this: if the image was truly from another world, perhaps what constitutes “good” composition — or even the fundamental reality of how things are perceived — might be radically different. Shouldn’t an app premised on showing glimpses of different universes actually show things that are different from our conventional norms?

In later demos with professional designers, we explored these ideas further — the subjective nature of taste, the challenges of maintaining visual consistency once curation has occurred, how other design apps set expectations, and the broader need to help designers mature their understanding of AI generated aesthetics. I initially struggled with some of this feedback, especially when it touched on taste and “good” style, areas often considered deeply personal and subjective.

The path forward became clear when discussing the app and feedback in another demo. The “a-ha!” moment solidified: the power wasn’t just in generating diverse outputs, but in using the app to explicitly provoke, to challenge taste, to question style, and to disrupt the status quo.

These feedback sessions led to significant updates: changing how results were displayed to emphasize difference, explicitly adding provocations challenging style, taste, and conventional composition. These provocations included statements like: “Dare to redefine your creative boundaries”, “Unlearn design. Relearn creativity”, “Your taste is a starting point, not a destination”, “Let your intuition guide you. Abandon the formula” and “Experiment without fear. There are no mistakes.” And perhaps the most exciting update was adding critiques for the visuals. These critiques, generated with AI, provided a vocabulary for aspiring designers, offering perspectives on why an image might be interesting, even if it defied traditional norms.

In total, only a few days of focused effort went into building the functional core of the app. I spent significantly more time writing about the app and creating a visual essay about the effort! This speed of creation, enabled by AI collaborations and vibe coding, was critical because it provided a tangible artifact to gather feedback against.

Story as the Core of Connection

The Multiverse Machine is more than just a collection of images and words; it’s about story. The story of the multiverse has played a significant role in its development, demos, and in detailing the research. Through thiskernel of an idea, I’ve been able to introduce ideas that are ridiculous, ideas that sound familiar — but upon scrutiny are unfamiliar, and found ways to connect to learning and literacy that are grounded in our day-to-day. And, my AI colleagues played a key role in writing the story, adding to the story, and refining the story. Story has made it possible to ask people to suspend their disbelief, to look at things differently, to ask “…how might we?”, and consider that the absurd and the familiar are not all that far apart. It’s fascinating how AI, despite its non-human nature, can help advance these inherently human stories, fostering deeper connections.

A more direct approach at tackling the challenge of creation, seeing, and doing might have been both less interesting and less effective. An approach more focused on the technology might have led the reader and user astray in other directions — focused on “…how might we optimize the code?” or “…did the AI generate too many fingers on that hand?” In both cases, the lack of story could leave us missing out on the opportunity to think differently, see differently, and create differently.

“Good” in this context isn’t just about aesthetic perfection; it’s about the impact it has — its ability to inspire exploration, spark inspiration, and — hopefully — facilitate the telling of new stories. Getting to “good” with this project wasn’t a linear path defined by pre-set quality gates. It was about impact, story, exploration, and finding ways to collaborate effectively with AI as a team. It required actively seeking feedback, having a tangible asset (the app) to demo, and possessing the capability to iterate, adjust, improve, and test rapidly. The tools and the vibe coding approach empowered this workflow, allowing the app to get “better” (by its definition of success) and for me, as the creator and researcher, to get better as well.

Section 3: Why Legacy Definitions Fail Us Now & A New Framework

The hands-on experience with the Multiverse Machine-its rapid prototyping, iterative feedback loops with AI and human collaborators, and explicit provocations on taste and style-revealed a crucial truth that underpins our challenge: the traditional, static definitions of ‘good’ we’ve inherited are fundamentally inadequate for the dynamic, AI-driven landscape. This project wasn’t just about building an app; it was a living case study demonstrating why success in the AI era demands a departure from outdated metrics and an embrace of a more fluid, contextual understanding of value.

In recent days, as I scan my news feed, I see many articles about vibe coding and creating with AI. Often, they offer tips, tricks, and best practices for using the tools. But the tools are moving so fast — what’s a “best practice” today might be obsolete tomorrow. What I’m not seeing are indications that definitions of “good” are moving at the same pace.

Perhaps clinging to old views on “good” is more than just nostalgia; perhaps it’s a way to resist change, to try to hold back the tide of disruption that AI represents. And if we plod along, tethered to old definitions of “good” that prioritize rigidity, control, and predictable outcomes, perhaps it saves us from having to address the uncomfortable realities of the AI era. We can just kick the can of AI Debt down the road, avoiding the need to address governance debt because “good” is defined in ways that preclude us from moving fast or empowering distributed creation. Responsible AI and ethical debt? Again, no need to worry — that’s a future problem if we don’t adopt AI in ways that require immediate ethical consideration!

We need to challenge this inertia. We can look back and say: the old definitions of “good” certainly addressed some desired dimensions of creation — functionality, reliability, perhaps even some aspects of performance or usability (drawing from models like FURPS+). But they understandably weren’t designed for dimensions that have become critical in our current landscape. They emerged before widespread human-AI collaboration and couldn’t anticipate the need for a world where creation is collaborative with non-human entities. They addressed safety and privacy through technical lenses-reliability and security-but weren’t designed for the broader, human-centered considerations now essential: algorithmic bias, data consent, ongoing stakeholder discovery, and the social implications of AI-generated content. Perhaps the limited ability to incorporate change corresponds to limited engagement with stakeholders? Perhaps limiting engagement was seen to maintain control and avoid the complexity of diverse needs — at least to some later point long down the road?

This isn’t to say the industry hasn’t tried to evolve. Various ISO standards and corporate responsible AI frameworks have emerged, but they remain fragmented-either too technical for creators or too high-level for practical application. My recent research on agentic AI systems confirms this gap, showing that traditional control-focused approaches often hinder effective AI implementation and compromise both responsible AI goals and ROI (Ackerman, 2025). What’s missing is a framework designed specifically for the dynamic, iterative nature of human-AI collaboration.

Given this gap in existing frameworks, this redefinition of ‘good’ in the AI era can be understood through The Good Vibes Loop — a dynamic, iterative framework for ensuring relevance and impact. This loop is predicated on continuously sensing, adapting, and creating value by understanding that ‘good’ is no longer a static, universal constant, but a contextual, evolving target.

At its core, The Good Vibes Loop operates through:

• Establish Responsible Foundations: Integrate critical considerations like safety, privacy, and ethics from the outset, ensuring that the pursuit of ‘good’ is always responsible and aligned with societal values.
• Prioritize Human Connection: Prioritize centering the creation around story, media, and other creative works that foster deeper human connection, meaning, and emotional resonance.
• Engage AI as Creative Partners: Leverage AI tools and capabilities as integral team members to generate diverse outputs, accelerate iteration, and explore possibilities.
• Create Through Experimentation: Cultivate a culture where creations are hypotheses, explicitly designing and implementing feedback structures and rapid testing cycles for continuous evaluation.
• Gather Diverse Feedback: Actively engage a broad spectrum of human & AI stakeholders and personas, acknowledging that ‘good’ is defined by their varied needs, perspectives, and contexts.
• Reflect and Adapt: Systematically process all gathered feedback to derive actionable insights, fostering continuous learning and driving necessary adaptations for the next iteration of creation.

A simple model to guide us toward “good”

The Multiverse Machine project itself serves as a tangible illustration of The Good Vibes Loop in action. From its inception, the project aimed to Establish Responsible Foundations by advocating for thoughtful engagement and critical literacy, pushing boundaries on what “good” means in AI-driven creativity rather than unexamined adoption. It inherently worked to Prioritize Human Connection by centering on story, using the app to provoke thought, challenge norms, and facilitate new narratives for literacy in design. The core of its development was to Engage AI as Creative Partners , leveraging AI tools (Gemini, ImageFX, RunwayML, Firefly, and Bolt.new) to rapidly generate myriad visuals and conceptual elements, from the initial paragraph idea to functional worlds. This dynamic collaboration allowed the team to Create Through Experimentation , with creations viewed as hypotheses that led to significant updates like altered display methods and explicit provocations on taste. The app’s evolution was driven by Gather Diverse Feedback , actively engaging both human stakeholders during demos (creatives, professional designers, and mentors) and utilizing AI-driven critiques that provided new vocabulary and perspectives on AI-generated aesthetics. All this input then fueled the process to Reflect and Adapt, systematically processing the gathered feedback to derive actionable insights and drive necessary adaptations for the next iteration of creation.

This redefinition of “good” is deeply intertwined with the principles found in methodologies like “Four Steps to the Epiphany,” Lean Startup, and Google Design Sprints. These frameworks emphasize that true value comes from validated learning, not just building. They shift the focus from rushing to solutioning to diligently engaging in customer development and problem identification. The very term “problem-solving” often implicitly glosses over the critical front-end work of rigorous problem identification.

In the race to solution, we delay feedback, we shortchange learning, and we waste time and money on building the wrong things. The beauty of rapid creation tools and methodologies like vibe coding, especially when augmented by AI, is that they allow us to quickly create exploratory artifacts that serve as concrete views to study and react to. This provides tangible artifacts that enable us to rapidly test whether we are meeting the evolving standard of “good” in real-time, with the very people we aim to serve, as we validate both the problem space and then the solution space. This approach directly supports the continuous learning and adaptation emphasized by The Good Vibes Loop. AI, in this context, isn’t just about generating faster code or more images; it’s about accelerating the learning loop that ensures we are building value by solving the right problems.

Section 4: Navigating the Shift: Pitfalls & Opportunities

As we navigate this accelerating pace of AI-driven creation, it becomes even more imperative to proactively address the potential pitfalls of unexamined progress. The shift away from legacy definitions of ‘good’ is not an invitation to abandon established principles of responsibility; rather, it demands a heightened vigilance around safety, privacy, and ethical implications. True ‘goodness’ in this new landscape stems from a commitment to mindful innovation, where rapid creation is balanced with a deep understanding of downstream consequences and adherence to evolving policies and standards.

This vigilance, far from being a constraint, becomes the foundation upon which new and valuable opportunities are built.

If we are to learn from the Multiverse Machine experience, aside from the speed and the value of feedback, perhaps the most important takeaway lies in the provocations about “good,” style, and taste.

We need to actively challenge the status quo. Context matters immensely. What defines “good” for a rapid prototype exploring novel concepts is different from what defines “good” for a mission-critical enterprise system. We need to take a step back and evaluate what truly matters today, for you, for your users, and for the specific context of the creation. That should shape how we start to define “good.”

This article has argued that defining ‘good’ in the AI era demands moving beyond fixed definitions, embracing instead a creative intelligence that understands context, welcomes provocation, and fosters human-AI collaboration for impact. Building on these insights, the final article in this series, ‘Cultivating the AI-Native Mindset: Building Literacy Through Doing and Discovery,’ will explore practical steps for individuals and organizations to build the capabilities needed for the future of creation. Follow along to complete this exploration.

References

Ackerman, L. (2025). Perceptions of Agentic AI in Organizations: Implications for Responsible AI and ROI . arXiv preprint arXiv:2504.11564. https://arxiv.org/abs/2504.11564

Originally published at https://www.linkedin.com.