Generative AI for Leaders is available now in hardback, paperback, and Kindle.

—

Leaders don’t often have complete data to help them determine the best course of action in times of decision. To bridge the gap, they apply their intuition on top of what they do know. They might call this going by instinct, trusting their gut, reading the tea leaves, divine inspiration, etc. However they choose to describe or deploy it, it is a critical coping method for generating an intelligent response in the fog of the moment, like a calculated shortcut to an answer that feels true, even if it’s just an educated guess at the truth. 

Exceptional business leaders become masterful at deploying their intuitive powers. Steve Jobs observed in his biography, “Intuition is a very powerful thing, more powerful than intellect, in my opinion.” Jack Welch famously believed in trusting his intuition: “You’ve seen something so many times over your life or career that you just get what’s going on without a lot of deep thinking. Gut instinct is a deep, even subconscious, familiarity—the voice inside you that tells you, “Go for it now” or “No way—not ever.” 

Psychologists tell us that intuition springs from our human powers of pattern-matching as we mentally sort through our lived experiences to look for similar situations that apply to our present or future decision points. Our brains train on our unique life history, so new moments in time will be seen through the lens of those older experiences to help us put them into context faster to better learn from them.

Intuition works like muscle memory to:

• instantly recognize the type of problem we’re solving in context with the data on hand;
• the worthiness of a proposed solution to address it wholly or partly;
• and the probable outcomes associated with going one direction vs. another

Here’s the thing, though. In practice, generative AI works a lot like human intuition. 

Pattern matching: Parallels between generative AI and intuition

Developing an Intuitive Understanding of generative AI is one of the first concepts that SparkCognition’s founder and CEO Amir Husain lays out in his new book, Generative AI for Leaders, available now on Amazon. 

From the introduction: “In a sea of data, generative models find not just patterns but the underlying distribution that governs the data. They utilize this knowledge to generate new instances of data that share the characteristics of the original dataset. Thus, a generative model trained on a dataset of paintings can produce new images that echo the artistic styles it has learned.”

That explanation of how generative AI models work sounds a lot like intuition, doesn’t it? 

And this brings us full circle. Generative AI has rapidly become one of the most discussed tech trends in the world. Yet, it’s evolving so fast that most industry leaders struggle to define its boundaries and forecast its impacts. At the same time, they might be feeling the pressure to become first-movers in this space. So this is the classic problem—how do we get ahead of the curve when we lack full information? 

Answer: we have to intuit what is happening in real time and judge how to proceed. We need to gather enough information to do our pattern-matching exercise. Generative AI for Leaders helps leaders who may be short on extensive professional history with artificial intelligence or generative AI begin to develop their own intuitive understanding of:

• How generative AI really works
• Why generative AI is rapidly becoming so important in industry and government
• What steps should leaders be taking now to leverage it successfully in their business

In Chapter 2, Husain dives deep into the comparison between intuition and generative AI, likening the mental models we use daily to how LLMs recognize when and how a series of events forms and reliably predict the next link in the chain. 

Here’s an excerpt:

Developing an Intuitive Understanding

Let’s unpack how Generative AI works. Take, for instance, GPT-3. It operates by completing a phrase like “Mary had a little…” Most people would complete that phrase by saying “Mary had a little lamb” because that line is from a famous children’s nursery rhyme. Essentially, because a person has heard it before, they possess a mental model that can predict the next word when given a partial sentence.

This principle underlies the workings of large language models, a type of Generative AI. Models like GPT-3 and GPT-4, predict the next word in a sequence. However, before making this prediction, these models compute Word Embeddings, which capture the relationships among words. They consider not just a few words, but dozens and even hundreds. This focus on relationships allows the models to extract the rules of grammar, semantics, and more, going beyond merely predicting the next word.

Is Everything a Sequence?

Consider that a word is simply a token or an element in a sequence. A sentence defines a sequence, and a word may be the next token in that sequence. This idea extends to pixels—the small dots of color that make up an image. Like words, pixels can also be predicted in a sequence, with the color and intensity of the pixel being items a user wants to predict or generate. 

The concept of prediction doesn’t stop at pixels and words. Humans can also predict actions. For instance, if we see a curvature in the arm of a baseball pitcher along with the ball in his hand, we can generally predict what will happen next. He’s about to throw the ball. What’s intriguing is how much human activity consists of one action following another and how much human intelligence can be encapsulated by understanding the relationship of actions, tokens, words, or pixels among themselves, and then being able to predict the next one. If we step back and think about it, nearly every move we make, every moment of every single day, could be predicted by looking at what precedes it. If we break it down enough, is every single thing we do part of a sequence?

If everything is a sequence, are we looking at a future where workers just need to feed LLMs a partial sequence and let it do the rest? Not necessarily. In Chapter 2—“The Benefits of Generative AI”—Husain examines how this technology can increase productivity, improve decision-making, and more, but also points out that humans need to be responsible for the final results. It’s a game-changer of a tool, but the outcomes depend on you.

For example, talking about creativity in problem-solving, Husain writes about the force-multiplying effects of ideation at machine speed:

Generative AI can propose a variety of potential solutions for a given problem or task. This creative problem-solving capability can aid in decision-making by providing diverse options and perspectives, fostering innovative thinking, and challenging conventional wisdom. Think of those brainstorming sessions guided by a consultant costing thousands of dollars an hour. Where do they all start? By people placing their ideas on the board or on sticky notes placed on the wall.

That type of ideation comes at machine speed with LLMs. For instance, a Generative AI model can propose multiple strategic options for business expansion, each with its pros and cons based on data analysis, allowing leaders to weigh different approaches and make the best decision.

Generative AI can also contribute to collective decision-making processes by facilitating collaboration and consensus-building. AI models can collate and analyze diverse viewpoints, identify commonalities and differences, and generate a range of balanced, data-informed proposals for consideration. This can help to streamline decision-making in large, complex organizations, making it more inclusive, transparent, and effective. If technology is used the right way, it can help mitigate the impact of cognitive biases in decision-making. Humans are prone to a range of cognitive biases that can distort their judgment and decision-making. Generative AI models, if properly designed and trained, can provide a more objective, data-driven perspective, helping to counteract these biases and improve the quality of decisions.

Generative AI definitely has the potential to enhance decision-making, but like everything else, it is not a magic bullet. The quality of AI-generated insights and decisions remains heavily dependent on the quality of the data it learns from, the fine-tuning processes it is subjected to, and the prompts used to create output. It is crucial for organizations to ensure their data is accurate, comprehensive, and unbiased.

Decisions made by Generative AI should, for now, be subject to human oversight and judgment. AI models, for all their sophistication, don’t possess human intuition, ethical judgment, or contextual understanding. There remains a non-zero risk that over-reliance on AI could lead to poor decisions or unintended consequences. Ultimately, all organizations should be held responsible for the technology they deploy, including the AI they develop or anything they develop with AI.

Why you should read this book

Developing an intuitive understanding is an integral skill we all rely on to navigate through life. All humans do it, and until recently, it’s been thought of as exclusively human: while modern computers can crunch numbers better than us, we humans have been better at filling in the blanks where data falls short. 

But now, with generative AI, is that changing? If so, how fast, and to what end? 

Generative AI for Leaders is the first book written purposely for executive decision-makers to help understand these questions. And the fact is we don’t yet know all the answers. We must (you guessed it) develop an intuitive understanding of generative AI first. 

Amir Husain’s book is a must-read resource to find the critical concepts, technological principles, clear opportunities, and ethical considerations that leaders need to be aware of now to help their organizations take advantage of generative AI in the future. 

Get your copy today! Generative AI for Leaders is available in hardback, paperback, and Kindle.

The post Generative AI for Leaders: An Intuitive Understanding appeared first on SparkCognition.