正在学习

Embracing a Human-Centered Approach (5)

Embracing a Human-Centered Approach (5)

Your human brain is beautiful and astounding. When you raise your arm and catch a ball, your brain has processed a flood of complex visual information to calculate the ball’s trajectory, speed, and position. It has not only tapped into the capabilities of specialized regions of your brain to detect orientation and motion, but has also pulled in top-down processing that leverages your prior experience and the context of this specific moment to guide and refine your perception of the ball’s movement. And this all happens in a split second.

We can recognize and distinguish between thousands of faces, often with just a glimpse. Our brains even provide us with specialized neural circuits and a range of fine-tuned regions, often completely outside of our conscious awareness, that rapidly detect and process nuanced cues such as facial expressions, tone of voice, and body language to identify and respond to the emotions of others.

Our human brains are creative and imaginative powerhouses, pulling inspiration from our lived experiences, emotions, and subjective perspectives to create or invent new things. We continually make novel connections between concepts and ideas or imagine scenarios that have no grounding in reality. This complex cognitive process fuels artistic expression, scientific breakthroughs, and the discovery of new tools and technology in a way that is unlike any machine. As author and journalist David Brooks beautifully wrote, the human mind “evolved to love and bond with others; to seek the kind of wisdom that is held in the body; to physically navigate within nature and avoid the dangers therein; to pursue goodness; to marvel at and create beauty; to seek and create meaning.” And it does it all using only the same amount of energy it takes to power a dim light bulb.

Yet, our human brains also carry flaws and limitations. All of us carry artifacts from our ancient ancestors that function poorly in our modern world.

Cognitive biases, which lead to discrimination and irrational judgments, originated as mental shortcuts to support the quick decision-making essential to basic survival in the resource-limited world of early Homo sapiens. Our brains are constantly awash in chemistry—continually fluctuating hormones and neurotransmitters—that can impact our cognitive performance and ability to make smart decisions. Fight-or-flight responses, helpful when facing lions, tigers, and bears, can be triggered by modern banalities such as a reckless driver cutting us off in traffic and lead to a cascade of physiological changes that impact our mental abilities. Stress, anxiety, inadequate nutrition, and sleep deprivation—unfortunately prevalent in our modern world—diminish our attention span and problem-solving abilities.

Each of us also carries a unique recipe of cognitive weaknesses. We’ve all been on our own individual and often lifelong journeys to discover strategies and approaches that work best for our unique cognitive profile. But no matter how much we’ve succeeded, we all have areas where we could benefit from a cognitive boost. Some of these challenges are small, but many are quite difficult to overcome. Executive functioning problems, such as difficulties with planning, organizing, and managing time are widespread. Diagnoses of attention deficit hyperactivity disorder (ADHD), which can cause difficulties with attention and focus, are increasing. According to the Yale Center for Dyslexia and Creativity, an estimated one in five Americans has dyslexia.

Despite these challenges and limitations, our remarkable human brains continue to adapt and evolve. And now, with the advent of a new era of AI, we have powerful new tools that can open up exciting possibilities to enhance our mental capabilities in ways we’re only beginning to explore.

AI Gives Us New Tools

Thinking with AI puts our human needs and dreams at the center. We’re still relying on the almost magical abilities of our extraordinary human brain to drive us forward. But we are leveraging the new machine capabilities to help us.

At the core are AI models, such as large language models (LLMs), which are mathematical systems trained on vast repositories of the world’s digital data—from written language to videos, images, and charts. A handful of tech giants and startups have built highly advanced “frontier” models, including LLMs like GPT. However, we’re also seeing new kinds of models emerge. These include smaller models that enable AI capabilities on your phone without network connectivity, and specialized models tailored to specific industries or functional domains.

We use software to implement and interact with these models in real-world applications. In fact, AI models are increasingly woven into our everyday software and digital experiences—you may have noticed options to write, create, search, or edit with AI popping up in your enterprise software or social media apps. But we can also access the capabilities of the most advanced models directly through chatbots provided by the companies that built them. ChatGPT, developed by OpenAI, is one such chatbot. It offers access to OpenAI’s sophisticated GPT models through a simple interface that requires no technical knowledge.

Like our brains, AI models have limitations and flaws. To use AI effectively to enhance our thinking, we need to understand these limitations. I’ll explain these tradeoffs throughout this book so you can make informed decisions about when and how to use AI.

Unlock a New Possible with Machine + Human Intelligence

Thinking with AI means understanding how to bring machine and human intelligence together to achieve something that wasn’t possible before. We combine these two powerful forms of intelligence. Our natural brain functioning—what scholar Michael Ignatieff describes as “a distinctively, incorrigibly human activity that is a complex combination of conscious and unconscious, rational and intuitive, logical and emotional reflection . . . so complex that neither neurologists nor philosophers have found a way to model it”11—works alongside AI’s vast processing power to achieve what neither could do alone. It’s a collaboration with the potential to enhance our performance as individuals, teams, and entire organizations.

Recent research offers a glimpse into one facet of AI’s potential. A study of 100,000 people across eleven occupations in Denmark found that workers estimated that using ChatGPT could halve their working time for over thirty percent of their job tasks.12 However, the promise of AI extends far beyond the relatively mundane benefit of time savings. Those who master the art of collaborating with AI use the technology as a force multiplier for their human capability—to help them push the boundaries of what’s possible as a human. Here are some of the ways people with high AIQ can elevate their innate potential:

练习题

When catching a ball, the brain primarily uses which of the following to calculate the ball's trajectory?

A. Only visual memory from past experiences
B. Specialized regions for orientation and motion, combined with top-down processing
C. Random guesses based on the ball's color
D. Auditory cues from the sound of the ball moving

Which of the following is a key factor in the brain's ability to recognize and distinguish between thousands of faces?

A. Conscious analysis of facial features
B. Specialized neural circuits and fine-tuned regions outside conscious awareness
C. Memorization of facial names
D. Visual comparison with written descriptions

What are some ways the human brain demonstrates creativity and imagination? (Select all that apply)

A. Making novel connections between concepts
B. Imagining scenarios with no grounding in reality
C. Following strict logical rules
D. Fueling artistic expression and scientific breakthroughs
E. Relying solely on factual information

Cognitive biases originated as mental shortcuts to support quick decision-making in resource-limited environments.

Stress and sleep deprivation have no impact on cognitive performance.

According to the Yale Center for Dyslexia and Creativity, an estimated one in ___ Americans has dyslexia.

The human brain continues to adapt and evolve, and with the advent of AI, we have new tools to enhance our mental capabilities in ways we are only beginning to ___.

What are AI models, and how are they trained?

How do we interact with AI models in real-world applications?

What are some limitations of AI models? (Select all that apply)

A. They are perfect and have no flaws
B. They can be biased based on training data
C. They require no understanding of their limitations
D. They may struggle with tasks requiring human-like judgment
E. They are infallible in decision-making

Which of the following is a key motivation for a human-centered approach to AI?

A. To restrict AI access to a few elite individuals
B. To mitigate AI's risks through diverse perspectives
C. To eliminate the need for human creativity
D. To focus solely on technical advancements

What are some barriers to finding value in AI? (Select all that apply)

A. Uncertainty about how to start
B. Confusion caused by extreme narratives
C. Fear of not keeping up with the pace of change
D. Belief that AI is only for technical experts
E. Lack of interest in technology

When catching a ball, the human brain processes complex visual information using specialized regions and top-down processing. Which of the following AI concepts is analogous to this human cognitive process?

A. AI models are trained on vast repositories of data without any specific focus.
B. AI models, such as large language models (LLMs), are mathematical systems trained on specific types of data to perform specialized tasks.
C. AI models can only process data in isolation without leveraging prior information.
D. AI models are limited to processing only visual data.

The human brain has flaws and limitations, such as cognitive biases and the impact of stress on decision-making. Similarly, AI models have limitations. Which of the following are limitations of AI models?

A. AI models can be biased based on the data they are trained on.
B. AI models can only perform tasks they were explicitly programmed to do.
C. AI models can struggle with tasks outside their training data scope.
D. AI models are not affected by the quality of input data.
E. AI models can make incorrect predictions if the training data is not representative.

The human brain's ability to recognize and distinguish faces is similar to AI models' ability to classify and recognize patterns based on existing data.

Explain how the human brain's creativity and imagination can be compared to the capabilities of generative AI.

登录后解锁笔记、知识点解析、AI 问答

立即登录