📰 This article is adapted from The Creative Algorithm, my Substack newsletter about creativity, technology, and the human condition. Go check it out. It's free!

🗺️🗼🌉🗽⛲🧠

I have a love/hate relationship with academic papers—they can be overly long and laborious, inaccessibly complex, and use jargon that requires a master's degree to unpack. However, the information is more reliable because of peer scrutiny and institutional standards. More study, less story.

I also have a love/hate relationship with ‘sciency’ pop culture books—they can be overly long and glib, regressively simplistic, and every chapter opens with cherry-picked supporting anecdotes about ‘Dave’ or ‘Sarah.’ However, the information is more entertaining and memorable because there is some personality. Less study, more story.

Summaries are a useful way to separate the wheat from the chaff in academic and popular arenas. Still, they don’t lend themselves to thoughtful rumination or long-term retention. Summaries are the tourist brochures of the literary universe, the difference between vacationing in a country or looking at folded brochures of popular destinations.

This post is a stack of brochures about interesting academic papers and popular articles that caught my attention this week. If any of them pique your interest, I encourage you to backpack through these pieces in their entirety for a richer understanding of the subject matter.

1. Creativity and Artificial Intelligence

https://www.sciencedirect.com/science/article/pii/S0004370298000551

In her paper "Creativity and Artificial Intelligence," Margaret Boden concludes that although algorithms can be creative in novel ways (combinational, exploratory, and transformational), they still have significant hurdles to overcome related to the human experience.

Most Interesting Takeaways:

Understanding the Nature of Creativity: Creativity is a fundamental feature of human intelligence and encompasses the generation of new ideas, motivation, emotion, and cultural context. AI can assist creative professionals in generating new ideas but may face challenges in evaluating the creative output.

Types of Creativity: There are three main types of creativity: combinational, exploratory, and transformational.

• Combinational creativity involves novel combinations of familiar ideas
• Exploratory creativity focuses on generating novel ideas within a structured conceptual space
• Transformational creativity involves modifying the conceptual space to generate previously impossible ideas

AI Models of Creativity: AI models have made progress in exploring and generating creative outputs, particularly in the domain of exploratory creativity. These models rely on domain expertise, knowledge of specific conventions, and pre-defined conceptual spaces.

Challenges and Opportunities: AI-assisted creativity faces challenges in capturing the richness of human associative memory, expressing subjective values, and automating the evaluation of creative outputs. However, AI can still augment the creative process by exploring predefined conceptual spaces, providing novel ideas, and assisting in generating outputs within specific styles or genres.

AI should be seen as a tool to augment your creativity rather than replace it. Embrace the collaborative potential of AI and integrate it into your creative process, combining your expertise with AI's capabilities to explore, generate, and refine creative ideas.

2. Augmenting human innovation teams with artificial intelligence: Exploring transformer-based language models

https://onlinelibrary.wiley.com/doi/full/10.1111/jpim.12656

Bouschery, Blazevic, and Piller had my full attention when they mentioned the Double Diamond in their paper on augmenting human innovation teams with artificial intelligence. The Double Diamond is a design-thinking process with two discreet areas of diverging and converging thought: one that centres around the problem (“doing the right thing”) and another that focuses on the solution (“doing the thing right”).

Teams can harness the power of AI to augment their innovation abilities at every stage of the design thinking process.

Key Takeaways:

Understand the potential of AI: Transformer-based language models like GPT-3, can assist in various aspects of the creative process, including problem-solving, idea generation, sentiment analysis, and text summarization.

Embrace hybrid intelligence: AI should be viewed as a collaborator and team member rather than a replacement. By combining the strengths of human creativity and AI capabilities, you can achieve superior results and enhance your creative endeavours.

Use AI for problem articulation: Language models can help you extract essential knowledge and quickly identify relevant text passages. Use AI to summarize and distill complex information, enabling you to focus on applying acquired knowledge to your innovation tasks.

Leverage AI for knowledge expansion: Language models can access and integrate knowledge from diverse sources, even outside your area of expertise. This allows you to tap into a larger knowledge pool and uncover connections and insights that may have been missed.

• Example: Suppose you're working on an innovation project in the automotive industry. You can use a language model to quickly analyze and summarize research papers, industry reports, and customer feedback, extracting valuable insights about market trends, emerging technologies, and customer preferences.

Automate sentiment analysis: AI-powered language models excel in mining sentiments from large volumes of text, such as customer reviews or social media discussions. By automating sentiment analysis, you can gain valuable insights into customer needs, product features, and areas for improvement.

Leverage AI for idea generation: Language models can generate a continuous flow of ideas based on prompts or initial inputs. They can offer incremental and radical ideas, and you can adjust model parameters to control the scope and openness of the generated ideas.

3. Unlocking creative potential: Reappraising emotional events facilitates creativity for conventional thinkers

https://www.sciencedirect.com/science/article/abs/pii/S074959782200098X

Bauman and Young discovered that cognitive reappraisal, a strategy for reframing the meaning of emotional events, can boost creativity and open-mindedness in more rigid-minded thinkers. If you find yourself hosting an ideation session with conventional thinkers, try this little icebreaker called The Flip Side (something ChatGPT and I collaborated on).

The Flip Side: A Key for Closed Mindedness

Step 1 - Identify a Challenge: Ask each participant to write down a recent challenge or problem they faced in their work on a piece of paper. It should be something they found frustrating or difficult.

Step 2 - Share and Discuss: Have each participant share their challenge with the group. Encourage a brief discussion about why these situations were challenging and the emotions they evoked.

Step 3 - Flip the Challenge: Now, ask each participant to "flip" their challenge. That is, they should reframe the problem by finding a positive aspect or an opportunity for growth within it. For example, if the challenge was "I received negative feedback on my project," a flip could be "I received valuable insights on how to improve my project."

Step 4 - Share the Flip Side: Have each participant share their "flip side" with the group. Discuss how this new perspective changes their view of the situation and how it could lead to new, creative solutions.

Step 5 - Group Brainstorm: Now that everyone is in a more flexible, open-minded state, transition into the brainstorming session. Encourage participants to maintain this "flip side" mindset as they generate ideas.

4. How To Check If Something Was Written with AI

https://goldpenguin.org/blog/check-for-ai-content/

SPOILER ALERT: You can’t.

The article "How To Check If Something Was Written with AI," updated and maintained by Justin Gluska, discusses the rise of AI content creation and the challenges it presents, particularly in distinguishing AI-generated content from human-written content. It provides a comprehensive guide on how to detect if a piece of writing was created by an AI, offering various methods and tools to aid in this process including:

• Undetectable AI: A tool that leverages machine learning to spot AI-generated text, making it a handy ally in identifying content crafted by advanced AI models.
• Originality.ai: Not just a detector for AI-generated content, but also a plagiarism checker. It's a two-in-one tool that helps ensure your content is both original and authentic.
• Content at Scale AI Detector: This tool employs machine learning algorithms to detect AI-generated content. It's adept at identifying content created by a variety of AI models.
• Copyleaks AI Detector: Similar to Originality.ai, it's designed to catch both AI-generated content and plagiarism. A great tool for maintaining the uniqueness and human touch in your content.
• GLTR (Giant Language model Test Room): This tool, created by people from MIT-IBM Watson AI lab and Harvard NLP, uses colour to show the uniformity of the text and how likely a machine predicted the next word.
• Writer.com's AI Content Detector: Trained to spot content created by some of the most advanced AI models, this tool uses machine learning to detect AI-generated content.
• OpenAI Classifier: Developed by OpenAI, a leading organization in AI research, this tool leverages their extensive knowledge and experience to detect AI-generated content.

While all this effort is appreciated, every bit of copy I pasted from ChatGPT-4 fooled all of these tools into thinking the copy was of human origin. We’re already reaching the point where we can no longer ascertain the difference.

AI Prompt Engineering Isn’t the Future

https://hbr.org/2023/06/ai-prompt-engineering-isnt-the-future

Oguz Acar makes an interesting distinction for the Harvard Business Review: knowing how to write prompts is not the same as understanding how to solve problems. If you understand problem formulation (“the ability to identify, analyze, and delineate problems”), you'll be far better at directing an AI assistant to do what you want. Besides, you can already ask the AI to optimize the prompt for you.

I don’t completely agree, but after reading the article, I dug up some methodologies, tips, and tricks for each step of the Problem Formulation Technique.

Identify the Problem:

• Brainstorming: Gather a group of people and let them freely share their thoughts about what the problem might be. The goal is to generate as many ideas as possible.
• 5 Whys Technique: Ask "why" five times to get to the root cause of the problem. This can help you identify the real problem, not just its symptoms.

Understand the Problem:

• Research: Conduct thorough research to gather as much information as possible about the problem. This could involve reading relevant literature, consulting experts, or analyzing data.
• SWOT Analysis: Identify the Strengths, Weaknesses, Opportunities, and Threats related to the problem. This can help you understand the context of the problem.

Define the Problem:

• Challenge Statement: Write a clear, concise problem statement that summarizes the problem, who it affects, and the impact it has. This should be a simple, specific statement that everyone can understand. I especially like the How Might We… method.
• SMART Goals: Define what a successful solution would look like using SMART (Specific, Measurable, Achievable, Relevant, Time-bound) criteria.

Analyze the Problem:

• Fishbone Diagram (Ishikawa Diagram): This tool can help you visualize the causes of a problem and their sub-causes, helping you understand the complexity of the problem.
• Pareto Analysis: This technique can help you prioritize the issues that need to be addressed by identifying the causes of the problem that have the greatest impact.

Formulate the Problem:

• Reframing: Try to look at the problem from different perspectives (like the Flip Side technique above). This can help you see new possibilities and come up with innovative solutions.
• Gap Analysis: Identify the gap between the current state and the desired state. This can help you formulate the problem in terms of what needs to be done to bridge this gap.

You can still use AI to assist you with each one of these steps, but knowing the steps is the most important part. While I agree with the author in principle, I also believe that understanding how to craft and optimize your prompts will give you much stronger results. More on that later!

Enjoy the rest of the week. And happy hiking.