Creativity: A Skill of the Future 

Creativity synthesizes humans’ capacities for originality, inventiveness, problem-solving, flexibility and adaptability, knowledge integration, and value creation.

The importance of creativity as a universal human capability can’t be overstated. Creativity is a highly desirable trait—not only do creative individuals display higher levels of happiness and well-being but also increased abilities for complex problem-solving and adaptability.^10,16,17,18 In simple terms, creativity is the ability to come up with original ideas or solutions within novel or existing contexts. Therefore, it’s difficult to envision any situation in which creativity would not be instrumentally valuable in some way. 

Creativity synthesizes humans’ capacities for originality, inventiveness, problem-solving, flexibility and adaptability, knowledge integration, and value creation. When humans create something of value to them or others, it can enhance their sense of purpose, motivating them to continue expanding upon the value they’ve created in new and interesting ways. Consequently, it’s unsurprising that creative individuals tend to experience lower levels of stress, anxiety, and depression^5,6 alongside substantial increases in their mood and self-esteem.^7,9 Importantly, this can also result in higher levels of self-awareness and emotional resilience, which are highly advantageous attributes to possess when navigating the future of work.^{1,2,6,13,14,15} 

“It’s really important, especially in the day of technology, to emphasize the notion of humanity through creativity.” 

Self-awareness is integral to growth in a professional setting. Self-aware employees can identify their strengths and weaknesses with ease and clarity, meaning that they can dynamically improve on their weaknesses while simultaneously enhancing their strengths. In a work landscape where skill requirements are changing rapidly, this quality will become progressively more desirable. According to Paul Marca, VP of Learning and Dean of Ikigai Labs Academy, “It’s really important, especially in the day of technology, to emphasize the notion of humanity through creativity.” 

Moreover, emotionally resilient employees can cope with change and uncertainty—they aren’t easily discouraged by failure and are more willing to take risks as they continue to explore possible solutions to the problems they face. In doing so, they also cultivate a more open-minded perspective, which allows them to think flexibly under conditions of uncertainty, and Marca agrees that “ideation starts to build flexibility and adaptability.” Though self-awareness and emotional resilience can arise independently of creativity, individuals can actively develop these traits if they work towards becoming more creative. 

Overall, there are many psychological and professional benefits to creativity, which together, empower individuals to be more cognitively and emotionally adaptable and resilient. But, it doesn’t stop there—creativity can further boost professional performance by fostering innovation and enhancing productivity. 

Innovation constitutes the development of novel ideas or products that are useful—innovation can’t happen without creative ideation, though not all creative ideas are useful.⁴ Fortunately, creativity can also motivate methods and strategies that test the utility of ideas in a way that respects their novelty. On the other hand, creative individuals tend to display unorthodox tendencies, which can enhance their productivity. For instance, a certain strategy may work well, but this doesn’t indicate that it’s the best possible strategy. Creatives are unafraid to explore alternatives to their strategies, and while this might result in some short-term failures, ultimately, creative people tend to be more robust and persistent problem solvers. 

Human creativity is unlikely to be replaced by AI in the short-term, however, like many other human skill sets, it can certainly be augmented—as Connor Wright, Partnerships Manager at the Montreal AI Ethics Institute implies, “it [AI] certainly adds to human intelligence in the form of allowing us to do things that we perhaps wouldn’t have done otherwise.” Individuals who possess a basic degree of AI literacy, especially in terms of generative AI (GenAI), could leverage AI to increase their creative capacity—AI literate professionals are more likely to recognize the value of AI as a “thought partner,” and therefore, more likely to experiment and collaborate with AI to derive novel solutions and ideas in response to existing or future challenges. 

Leveraging Generative AI for Design Thinking

“Design thinking teaches you to ask questions rather than make statements.”

Design thinking is distinctly valuable in a business setting. A 2018 report by McKinsey¹² revealed that design-led businesses not only doubled their rate of investor return and profits but also, over a decade, outperformed the S&P 500 by a whopping 219%. In essence, design thinking is an approach that organizations can utilize to increase the flexibility and adaptability of their problem-solving strategies, to ensure that they continue to maximize their impact and shareholder value, even under conditions of constant change and innovation. In a nutshell, “Design thinking teaches you to ask questions rather than make statements,” says Marca, instilling a flexible and adaptable mindset that supports better change management.  

Therefore, we can only expect the value of design thinking to increase as AI innovations continue to proliferate and advance, causing profound changes to the work landscape, and unpredictably shifting the needs of society as a whole. According to the Interaction Design Foundation,¹¹ design thinking can be broken down into five stages: 

Stage 1: Empathy: Comprehending the needs, desires, and experiences of those for whom you are designing. This necessitates engaging with users and stakeholders to gain detailed and honest insights into their perspectives and challenges.

Stage 2: Define: Defining the problem you’re addressing with respect to the insights you’ve gathered. This requires synthesizing observations about your target audience and identifying the core problems they face.

Stage 3: Ideate: Generating a diverse range of ideas and potential solutions once you’ve defined the relevant challenges or problems your audience faces. This stage encourages creative thinking and exploration of a wide array of possibilities without immediate judgment or limitations.

Stage 4: Prototype: Developing tangible representations or models of one or more of your ideas. These prototypes can be simple and low-fidelity, aimed at solidifying ideas and facilitating further exploration and testing.

Stage 5: Test: Engaging your target audience and stakeholders with your prototypes to gather feedback, learn what works and what doesn’t, and understand how the solution might be improved.

Novel problems typically require novel solutions. Even though creativity is emphasized in stage 3 of the design thinking process, it can permeate all stages from empathy to testing. For example, people’s needs frequently fluctuate, and to figure out what their true needs are, we may need to creatively explore alternative approaches to stakeholder engagement that indicate how needs might evolve and materialize in the future. Similarly, a novel idea or product may be difficult to test within an existing framework, and may therefore require an original testing approach. 

“Ask it [AI] how it got from point A to point B, so that one can learn about how it has, with all this information, been able to piece together these ideas.”

Fortunately, the advent of GenAI provides individuals with a powerful tool to enhance their design thinking capabilities, substantially increasing the value they provide throughout a work landscape that’s saturated with uncertainty. For instance, during stage 1, users could leverage AI to streamline information gathering and data on their target audience, which they can then use to inform and develop the questions they intend to ask, resulting in a series of targeted questions that reveal actionable insights more easily. However, as Stephen Smith, CEO and Co-Founder of SimplyPut AI recognizes, “There’s always this push and pull between the deterministic nature of AI and the ability to take advantage of its creativity,” and as fellow Co-Founder Tony Doran adds, 

“We [SimplyPut AI] have the AI citing the examples that the data people actually trust and can offer that citation to the end user so that the end user isn’t only seeing the creative AI output, but also the examples that the data people have blessed.”

During stage 2, users could input all of their observational data into an AI model, and then proceed to prompt the model to summarize and identify the key commonalities or trends that emerge. From here, users could iteratively refine and formalize these commonalities into concrete problems or challenges that their target audience faces. Still, it’s important that users, “always look for signals of why something [an AI-generated output] might be true—prompt the AI to show why it knows what it knows,” counsels Doran, and fellow Co-Founder, Smith echoes Doran’s words, advising that users, “ask it [AI] how it got from point A to point B, so that one can learn about how it has, with all this information, been able to piece together these ideas.”

“Most of us think in terms of tens or hundreds of ideas, but if you’re really stretching, you need to think much more broadly—a great way to do that is to build a team. But, if you’re an independent person trying to figure out a pathway, using a tool like ChatGPT to generate those ideas is really critical.”

During stage 3, users could leverage the vast stores of human knowledge contained within AI models to expand the range of ideas and potential solutions linked to the previously identified challenges. AI models can help users identify solutions that have proved successful in the past, or alternatively, synthesize an array of ideas to develop a novel problem-solving approach. Moreover, users can prompt AI to challenge the ideas they come up with and provide better alternatives, which although not always immediately useful, can help cultivate a more creative, diverse, and explorative ideation process. 

“Most of us think in terms of tens or hundreds of ideas, but if you’re really stretching, you need to think much more broadly—a great way to do that is to build a team. But, if you’re an independent person trying to figure out a pathway, using a tool like ChatGPT to generate those ideas is really critical,” states Marca. 

During stage 4, users can prompt AI models to generate a variety of possible prototypes that encapsulate and reflect the solution they’ve come up with. This process will require iterative refinement on behalf of users since it’s unlikely that the initial AI-generated prototypes will be adequate—as claimed by one F500 healthcare company executive, “I can’t look at ChatGPT or at any of the learning models out there and go, okay, I need to do X and I can do it in two steps or two questions to get to it—it has to be iterative.” For example, users may find an initial AI-generated prototype to be interesting but practically unfeasible. However, by isolating this prototype and prompting AI to generate several alternatives that each coincide with stated feasibility objectives, users can flesh out a prototype with real-world applicability and value. 

During stage 5, users can input their finalized prototype ideas into an AI model, prompting it to suggest a series of metrics that evaluate the performance of the prototype. Much like the previous stage, this will also require iterative refinement on behalf of users, since initial AI-generated metrics will offer a useful starting point, but not a concrete end-point. In other words, “It’s important for the humans—the user on the front-end—to be able to see and articulate and know what the dependencies are and what they’re trying to solve for,” recommends the F500 healthcare company executive. Moreover, recent developments in AI-assisted search capabilities could dramatically enhance this process, enabling users to prompt models to consider the most recent test cases for similar prototypes as they develop a testing framework. 

Now, what might all this look like when put into action, such as in the context of developing a product or service? We illustrate a brief example below of a recruitment company that wants to create a digital service that allows registered users to input their profiles and get matched with potential job opportunities (For the sake of this example, let’s imagine that platforms like Linkedin and ZipRecruiter don’t yet exist).  

Example: The company already has access to a wealth of data on the problems professionals face when trying to source new job opportunities, but it doesn’t have the time or resources to analyze it internally. Therefore, the company decides to input this data into a GenAI model, prompting it to analyze and extract relevant trends. Once trends are formalized, the company prompts the model to design a digital survey that considers the pain points their target audience faces in terms of the previously identified trends. After receiving a sufficient number of survey responses, the company synthesizes the survey results into one cohesive dataset, once more feeding it into the model. But this time, the company asks the model to conduct a sentiment analysis on survey data, to determine whether the responses provided align with the previously identified trends. Assuming that they do, the company begins developing its prototype job-matching algorithm, iteratively refining the code by leveraging GenAI coding tools. Once the prototype is finished, it undergoes an additional AI evaluation, whereby the AI is prompted to generate performance metrics that allow the company to monitor and measure algorithmic performance over time. 

This example could go on indefinitely, and therefore, we offer a challenge to readers that may prove to be a useful design thinking exercise: once a prototype becomes a concrete product or service, how might we continue to leverage AI to improve, adapt, or re-imagine it? 

All in all, leveraging GenAI to enhance design thinking will require a basic level of AI literacy, namely in terms of rudimentary prompt engineering skills and a general understanding of models’ limits and capabilities—for those who want to go the extra mile, they should explore the range of purpose-built tools that are suited to overcoming and addressing the challenges mentioned at each of these stages. 

Though there are many prompt engineering resources available to users, we suggest that they begin by simply experimenting with GenAI on their own to discover what works and what doesn’t—knowledge gained through experience tends to be both more practical and long-lasting. If you find yourself craving more detail, information, and guidance, there are three other essays in this series worth exploring. 

As for the limitations that advanced GenAI models may encounter, we’ll discuss these in detail at the end of this essay. But first, we explore some additional ways in which AI could empower human creativity. 

Leveraging Generative AI to Empower Creativity

In today’s world, many companies have resorted to utilizing crowdsourcing and innovation contests to generate novel, innovative ideas and solutions to the problems or challenges they encounter. While this tactic yields occasional success, it doesn’t guarantee that high-quality ideas and solutions will always emerge or materialize. There are a few reasons why this is the case. 

Harvard research⁸ findings have highlighted four important challenges that demonstrate why crowdsourcing and innovation contests may sometimes fall short: 

1. Crowdsourcing can lead to a massive influx of ideas, which can make evaluation extremely difficult. Merging ideas, identifying good ones, figuring out which combinations are best—these tasks are complex and tricky to execute.
2. Resorting to domain expertise doesn’t always resolve the problem—experts are really good at creating and identifying actionable ideas but they aren’t great at accepting or coming up with new ones. 
3. People without domain expertise may provide novel good ideas but lack the technical abilities or knowledge necessary to execute them. 
4. Companies may struggle to see the big picture—businesses must consider a host of individual requirements they need to fulfill, which can cause them to overlook critical collective requirements. For instance, a crowdsourcing initiative may yield a good solution to a problem, but enacting that solution might require that a company undermines a core value, such as user privacy. 

It may be obvious by now, but AI-augmented design thinking can offer tremendous value in this context. At each stage of the creative ideation process, from empathizing to testing, those who leverage GenAI will be less likely to succumb to the challenges mentioned above, primarily because they’ll be able to benefit from an iterative and collaborative human-AI interaction that continually challenges, refines, and improves upon novel ideas or solutions. In the words of Marca, “Once a decision has been made, it’s really about communication and the execution of a change project,” and AI will help us both streamline and deepen this process. 

In fact, the same Harvard study that noted the above challenges also identified five key benefits of AI in creative ideation, though we have slightly redefined and expanded upon these benefits to highlight certain critical components. Importantly, as you read through these, consider how they relate to AI-assisted design thinking and more broadly, the value of cultivating AI literacy. The benefits are listed below:  

• Idea association: Sometimes, people can intuitively spot a connection between two remote or seemingly unrelated ideas. However, formalizing or defining that connection in a way that makes sense to them and others can be challenging. GenAI is especially good at formalizing connections between disparate concepts, and illustrating them in a way that is comprehensible to a wide audience (or targeted audience if that’s what you’re looking for). This AI capability can also encourage users to discover value in remote places where others have been afraid or unwilling to venture to in the past. 

• Thinking beyond what is possible: When humans create things, they typically do so within the confines of what’s deemed possible, acceptable, relevant, useful, etc. Nonetheless, as we’ve seen with some of the greatest artists and inventors in human history, creativity can transcend possibility, and it’s often once it has done so that it provides the most value. When we leverage GenAI as a creative “thought partner,” we can literally prompt it to challenge every belief, preconception, solution, idea, or concept that we discuss. This kind of interaction can encourage us to think beyond our common understanding of form and function. 

• Measuring the value of ideas: Prompting GenAI models to compare and contrast ideas, to generate hypothetical scenarios in which ideas are tested, to provide perspectives from leading experts on the value of such ideas—these capabilities scratch the surface of what GenAI can do to help us evaluate our creative outputs. As mentioned earlier, GenAI can also help us identify and create metrics to evaluate our ideas, though users should be cautious in accepting initial metrics since they’ll likely require further refinement. 

• Idea synthesis: Though idea association and idea synthesis are connected, they aren’t exactly the same thing. Idea synthesis concerns GenAIs ability to not only find connections between disparate ideas but also synthesize a conglomeration of them to produce a holistic solution or concept. In essence, GenAI can help individuals build an ideological epicenter. 

• The cross-pollination of ideas: AI already influences team workflow and communication,³ usually via integration with existing collaborative platforms such as Slack, Notion, and Trello. This kind of AI integration can increase the frequency with which interdisciplinary teams collaborate, while real-time translation capabilities allow for the development of more diverse and globally representative teams. This could enable teams to more efficiently share their ideas internally and externally, gaining useful insights from team members and relevant experts who typically wouldn’t have been able to review ideas. 

GenAI can and will empower people to be more creative. However, to unlock and tap into this potential, individuals must be AI literate. Realizing the benefits of GenAI for creativity will require that people understand how to optimize their interactions with AI to produce high utility outputs—to do so, they should experiment with GenAI throughout as many different contexts and domains as possible. Through consistent experimentation, individuals can streamline their AI literacy learning curve, acquiring functional AI skills in addition to an understanding of AI’s capabilities repertoire, potential novel use cases, and related risks and impacts. 

“Often the most difficult thing to do as a writer, strategist, or design thinker is to come up with the initial idea, so having a tool like this [GenAI] will allow you to generate lots and lots of ideas.” 

Before concluding this section, it’s important to expand on the idea of AI as a “thought partner”. There are two additional angles to consider here, the first of which we’ve already briefly alluded to: 

• Users can prompt AI to assume the perspective of well-known experts on a variety of issues, even ones on which such experts haven’t produced any work. For instance, a user could ask, “If he were alive today, what solutions might Charles Darwin offer to combat climate change?” or “If the Founding Fathers were alive during the time of the Roman Empire, how might this have influenced the development of the US Constitution?” These kinds of questions can inspire users to take more creative viewpoints on the problems they face, and in many cases, apply the knowledge of some of the brightest minds in human history in real time. Importantly, this technique can also be applied to established argumentative viewpoints (e.g., how would existential philosophy tackle the question of extraterrestrial life?), beliefs (e.g., what might a conversation between a Priest, Buddhist, and Rabbi look like?), and theories (e.g., what might a unified theoretical framework between String Theory and Evolutionary Biology look like?).  

• Users can leverage AI to help them define, execute, or identify a creative strategy. The benefits of AI for creative ideation presuppose that the user already has an idea, but this isn’t always the case, even among those of us who are highly creative—to this point, Marca expresses that, “Often the most difficult thing to do as a writer, strategist, or design thinker is to come up with the initial idea, so having a tool like this [GenAI] will allow you to generate lots and lots of ideas.” 

For instance, a user may want to learn about design thinking but has no idea where to start. In this context, only a simple prompt is required: “Identify the most successful design thinking frameworks, and summarize their key concepts in detail.” If the user wishes to go a step further, they could prompt the AI to: “Using the previously identified design thinking frameworks, generate a novel idea for each framework, but ensure that in generating this idea, you follow the steps proposed by the framework.” The user can then compare the various AI-generated ideas that emerge, and in doing so, gain a better understanding of which design thinking framework is best suited to their needs. 

The Destruction of Value: Limitations to Advanced Generative AI Models 

People tend to place confidence in GenAI in areas where it lacks competence, whereas in areas where it’s very competent, people tend to distrust it.

Despite the impressive capabilities of GenAI models, they still exhibit several limitations. Interestingly, however, people tend to place confidence in GenAI in areas where it lacks competence, whereas in areas where it’s very competent, people tend to distrust it. Therefore, an understanding of GenAI limitations is instrumentally valuable in terms of identifying and optimizing AI capabilities. If people fail to comprehend GenAI’s limitations, they risk destroying or undermining the value of what they’ve created. 

We won’t dive into all the limitations of GenAI since that’s beyond the scope of this essay and requires a much deeper technical understanding of these systems that could classify as AI fluency rather than AI literacy. However, we do explore the most prevalent limitations below: 

• Algorithmic and data bias: GenAI has a tendency to produce biased outputs. Much of the time, this is due to biases in the data on which models were trained. However, statistical and algorithmic biases can also be a factor. 

• Information hallucination: GenAI models may output information that’s blatantly false or inaccurate, such as sources that don’t exist, or facts that have been proved incorrect. Importantly, models will always generate an output in response to a prompt, and this can persuade users to believe the output is legitimate. In this context, Wright agrees, in that “there’s always going to be an output and there’s always going to be an input—that’s what the system is designed to do, to provide the best possible output from the input that was received.” 

• Contextual and ambiguous understanding: GenAI models are pre-trained on large datasets. This gives them the ability to execute an array of tasks across known task domains. However, in novel or specific contexts that deviate from such task domains, the performance of GenAI models may be unreliable or inconsistent. Models may also struggle to perform under ambiguous or novel conditions due to generalizability limitations stemming from their training data. 

• Emotional and cultural intelligence: Although GenAI models are trained on monumental amounts of historical human data, they still lack emotional and cultural intelligence. Emotion and culture are nuanced characteristics of human existence, and the way they manifest themselves within each individual may differ. Though AI models may be able to internalize some common aspects of emotional and cultural intelligence, they still fall short in this respect. 

• Memory: GenAI models possess limited memory capacity—they struggle to recall and incorporate earlier parts of a conversation into the most recent answers they provide. Typically, a user prompt should include all the information that’s required for an AI model to answer the question or solve the problem. Frontier GenAI models have enormous context windows, so users shouldn’t hesitate to craft highly detailed and specific prompts. 

• Intuitive adaptation: Humans can intuitively adapt how they communicate with others based on social or behavioral cues and signals. Even though GenAI models may sometimes appear to behave in a convincingly “human” way, they can’t intuitively adapt themselves to user preferences, however, a lower degree of non-intuitive pattern recognition-based adaptation is possible.  

• Personalization: While certain GenAI models can be fine-tuned or personalized for specific task domains or individuals (e.g., a chatbot), they still lack the ability to understand individual human experiences and histories. Models can tailor their responses with respect to the inputs they receive in a given conversation window, but this represents the limit of their personalization capabilities. However, this is a major area of potential future growth in GenAI models, and subsequent iterations will likely possess much more sophisticated personalization capabilities.

• Common sense reasoning: Unlike humans, GenAI doesn’t possess common sense reasoning: an intuitive understanding of the physical and social characteristics that underlie the real world. Humans directly experience the world in its full complexity whereas GenAI models don’t yet have any phenomenological capacity.

• Dependency on user input: The outputs that GenAI models produce are only as good as the inputs that users provide. The more detailed, precise, and accurate an input is, the better the output will be, and vice versa. 

If users don’t consider these limitations as they interact with GenAI models, they risk undermining their abilities to maximize the value that such models can produce. These limitations also further emphasize the importance of AI literacy as an integral characteristic that shapes how we interact with and benefit from AI. 

Conclusion

The multifaceted nature of creativity, its integral role in design thinking, and the augmentation potential offered by GenAI models present a compelling case for the growing importance of creativity in the modern world. Creativity, far from being a mere artistic endeavor, emerges as a critical skill for problem-solving, innovation, and adaptation in a rapidly evolving work landscape. The psychological and professional benefits of creativity extend beyond individual fulfillment, contributing to emotional resilience, self-awareness, and a heightened capacity for complex problem-solving.

The intimate relationship between human creativity and GenAI, particularly in enhancing design thinking processes, underscores the transformative impact of AI on creative endeavors. GenAI can be a catalyst for innovative thinking, expanding the boundaries of what’s possible in ideation, problem definition, prototyping, and testing. 

However, the limitations of GenAI, including its biases, dependency on user input, and lack of emotional and cultural intelligence, highlight the fundamental value of human creativity. While AI can augment and streamline creative processes, the human element—our ability to empathize, intuit, and conceptualize beyond data—is what truly drives innovation and value creation, while also providing the judgment necessary to navigate AI’s limitations. More concretely, Doran also advises that individuals, “unlearn that your value isn’t necessarily your paid-for task, especially in a work context.” Overall, thinking creatively enables us to look beyond our ‘numeric value’.  

This exploration reveals that the future isn’t just about technological prowess but equally about nurturing and leveraging human creativity. As we move forward, the balance between these two notions will deeply influence our capacity to thrive in a world that’s constantly in flux. Developing AI literacy and harnessing the strengths of GenAI while also cultivating our creative skills, will be instrumental in shaping a future where technology and human creativity coexist in a mutually augmentative relationship.

References 

*note: references are ordered alphabetically by author name, with links provided where appropriate. 

1. The Psychology of Creativity and Resilience (Bar-Sawme, 2023) 

2. Enhancing Self-Awareness Through Creative Experiential-Learning Play-Based Activities (Bell et al., 2014) 

3. The Future of Work and Leadership in the Age of AI (Berkeley Executive Education, 2024)

4. The Importance of Creativity in Business (Boyles, 2022)

5. Creativity is Your Secret Advantage for Mental Health and Well-Being (Brenner, 2019)

6. Everyday creative activity as a path to flourishing (Connor, DeYoung & Silvia, 2016) 

7. The Mental Health Benefits of Creativity (Diversus Health, 2024) 

8. How Generative AI Can Augment Human Creativity (Eapen et al., 2023) 

9. How Creativity Positively Impacts Your Health (Field, 2021) 

10. Creative Occupations and Subjective Wellbeing (Fujiwara, Dolan & Lawton, 2015)

11. Design Thinking (Interaction Design Foundation, 2016) 

12. What is design thinking? (Mckinsey, 2023) 

13. Surprising Link Between Creativity and Resilience (Naiman, 2023) 

14. Plea for a Daily Creative Act (Riesen, 2020) 

15. Self Awareness Through the Creative Arts (Sanoff, 1971) 

16. Subjective well-being and its sub-scales among students: The study of role of creativity and self-efficacy (Tamannaeifar & Motaghedifard, 2014) 

17. Problem-Solving Ability and Stress Mediate the Relationship Between Creativity and Happiness (Tan et al., 2019) 

18. Being Creative Makes You Happier: The Positive Effect of Creativity on Subjective Well-Being (Tan et al., 2021)