Josh Bersin 认为 AI 无法在复杂决策中胜过人类直觉

We’ve now spent years debating the use of AI for human decision-making: who to hire, who to promote, how much to pay someone, and hundreds of other things. Leaders are faced with complex, difficult decisions every day: can we trust AI to make them for us?

I’d suggest the answer is no, and that’s the topic of my latest podcast.

What Is Intuition? And What Are Emotions?

We all know that “Type 1 Thinking,” or gut-feel reactions, dominate our mind every day. We meet someone or sit in a meeting, and the decision of “who to hire” or “what to do” suddenly becomes clear, even though the data is hard to find.

I’ve been doing a lot of research on genetics, emotions, and intuition and I’m coming to the conclusion that no amount of AI “superintelligence” can substitute or replace our emotions. And these emotions, which come from our history, upbringing, and genetic makeup, are often more insightful than data may say.

As an engineer, I’m a huge fan of data and science. So in no way can I argue that algorithmic and data-driven decisions are bad. But as much of my research in human capital has shown, it’s the human “gut-feel” that complements, aids, and finalizes the decisions that machines may make.

Here’s my technical argument. AI systems use probabilistic neural networking to “train” the model to think. The AI model then looks at new information and goes back to its training to answer a question, write a block of code, create a picture, or write an essay. It does a rather spectacular job of doing this because it can almost instantly look at its entire corpus as one giant “data set” and use vector calculus to answer.

This all assumes that the core data, which may be every document ever written, actually has enough insights or perspectives to make this decision. And most big AI labs now admit that we’ve “run out of data” to index, so they’re creating synthetic data (data created by AI from data) to enrich the model.

What’s Missing?

Well if you read the research on emotions (there are at least six theories), you see that most studies believe that “how we feel” about a situation is based on our lived experience, the nature of the stimulus (what we see, hear, feel), and our genetic makeup. And this last factor, our genes, takes us back to millions of year of human evolution.

So while a business decision may seem logical and data-driven, we each interpret data differently and our experience and human nature define our reaction. This is why, as I discuss in the podcast, a team of executives can all look at the same data and come to wildly different decisions about what to do next.

(The example I cite is a meeting that looks at revenue growth relative to market conditions and company operations, and one executive says “we should be proud of where we are and celebrate our success” and another claims “why aren’t we growing faster, we can do better than this!”)

As with every human interaction, some people are positive, creative, ambitious thinkers and others are conservative, slower, deliberate types. This magical combination of human “intuitions” is what makes some companies out-perform others in any competitive market.

Where does this come from? It comes from millions of years of evolution and our own unique set of epigenetic capabilities. In other words, our human intelligence (and emotional intuitions) are based on our own geneology, experiences, and history. My father and his lineage was musicians and scientists; my mother’s family were merchants and sales people. I wound up as an engineer, fascinated by business and people. And because both my parents were entrepreneurs, I am ambitious, risk-taking, and somewhat of a thrill seeker. (Read “The Gene” for a fascinating explanation.)

These human “skills,” as we may call them, come from millions of years of history and billions of gene combinations. And they manifest them in our minds and bodies as emotions, intuitions, strengths, weaknesses, and ultimately as wisdom.

Can some algorithmic form of “Superintelligence” possibly outperform this? No way.

AI vs. Human Decisions

Daniel Kahneman’s popular book “Thinking Fast and Slow” tries to simplify this as two “types” of intelligence: fast and slow. He argues that “fast” intelligence is intuition, and “slow” intelligence is analytic.

Well as popular as that book was, we can now see that things are far more complicated. AI systems can perform “slow” intelligence reasonably well, but they’re actually incredibly naive. Ask Grok to explain the Jeffrey Epstein story and it does a very poor job of explaining our human intuition that “it’s a sordid, messy, embarrassing and disgraceful affair.”

While AI is an ongoing phenomena, and companies are now investing hundreds of billions into data centers, my point is simply this. No amount of computing is going to replicate the genetic, historic, evolutional intelligence we have as humans.

When you drive up the street and see a child at the curb your “instincts” tell you she is going to run into your path. Can we “teach” a robot to think that way?

When you sit through a meeting and the “decision” we made feels incorrect, your “instincts” tell you it’s time to delay, and see how you feel tomorrow. The AI would jump to an answer and plow ahead.

These human, genetic, emotional, instinctive processing powers we have as humans are incredibly important. These are the things that differentiate Steve Jobs from Bill Gates; they explain the behaviors of Elon Musk vs. Sam Altman. We must respect these human, intuitive, emotional parts of our intelligence, they’re even more important today in the world of AI.

Technically speaking, the AI community defines Artificial General Intelligence (GAI) as “A type of artificial intelligence system with the ability to understand, learn, and apply knowledge across a wide range of tasks at the level of, or beyond, that of a typical human.” (ChatGPT definition here.)

Given the richness of our minds and emotions, I doubt this will happen any time soon.

Additional Information

We are in the middle of a major study of AI implementations in all areas of HR and business. If you’re learning something or want to debate these topics, please reach out to us or ask to join our Big Reset Groups. And if you’d like to debate the intricacies of human intelligence vs. machine intelligence, let me know and we’ll set up a call.

The Six Major Theories Of Emotion

The Rise of The Superworker

Understanding The Human Side Of Business: The Irresistible Organization

The Gene: An Intimate History, by Siddartha Mukhergee

ChatGPT’s Explains “How Does Human Intuition Compare to AI?”