“In the AI Era, Who Will Be Left Behind?” Harvard and McKinsey alumni Bradley and Harvey discuss this topic in the video “Decision-Making in Another Dimension: Pursuing Precision May Be Meaningless.”
Bradley first mentions that AI cannot “read the room,” but this is key to making things happen. In management consulting, the most time-consuming part is deliberating what content to include. For example, “Will this wording trigger emotional reactions in some people?” AI finds it difficult to judge such questions, which also tests humans’ grasp of the现场状况.
Harvey points out that trained, experienced individuals see AI as a powerful tool. Because they know what is “good,” they can refine AI outputs to improve them. But newcomers just out of school don’t know what’s good; they accept AI’s answers at face value. The future requires using AI to achieve 80 points and then fine-tuning to 100. But those lacking judgment will simply settle for 80. No matter how much AI helps, you need to add that final touch—only if you understand what’s missing.
AI cannot read the room, but that’s the key to grounding business.
Bradley states that AI cannot “read the room” yet. Currently, AI cannot truly understand possible emotional reactions. But this is crucial for “how to truly land an idea.” You still need to sit down and interact with people to grasp these subtle nuances. From a business perspective, this becomes a very critical and highly differentiating learning point.
After AI emerged, data analysis will inevitably be replaced gradually—an irreversible trend. But ultimately, it comes down to who can interpret these analyzed data well, make good judgments, and build consensus within the team to move forward.
This is the real challenge. Many judgments still require human interpretation.
The author adds: Large language models, as the name suggests, are trained by feeding vast amounts of data, enabling them to predict the most likely next words. They rely on huge databases to understand context and generate content. They know that phrases like “an apple falls” appear often, but they don’t understand gravity. Similarly, they struggle with abstract situations like “reading the room.” In fields like data analysis, they are only limited by model efficiency.
Using AI for reports? The real key is judgment.
Bradley uses presentations as an example. In academia, a lot of time is spent adjusting formats, analyzing data, and organizing content for presentations and discussions. But in management consulting, the real time-consuming part is “deliberating what content to include” and “how to phrase the same sentence.”
Harvey further explains that, on-site, certain words might be sensitive or trigger intuitive emotional reactions. Because once someone’s amygdala is triggered, they basically shut down.
AI is inherently limited and struggles with abstract, subjective judgments like “Will this wording trigger an emotional response?” So even if AI improves efficiency, the most important role still belongs to humans for overall situational judgment.
Here, “judgment” is the key word. Harvey says that when you can make these judgments, it indicates you understand the people on-site, their backgrounds, and what they’re thinking.
So every time someone says, “AI can do presentations now, so management consultants will disappear,” Harvey responds that if it’s just about delivering analysis reports, AI can definitely replace you. But if the goal of the presentation is “to facilitate change,” then judgment is critical. If you let AI fully represent you, you also take on a risk: its value judgments become yours. But what exactly are AI’s value judgments? To some extent, they remain a black box.
The essence of PowerPoint is to make a point.
Harvey highlights a key point: a presentation isn’t just about completing a task; it’s about “making a point.” PowerPoint is called PowerPoint because it helps you convey your viewpoint. You need presentations to communicate ideas, but that’s not enough—you still need other skills.
He mentions that a manager’s role, to some extent, is to leverage their perspective—what counts as good? When should they send things back for rework? How to handle people? These are the most important leadership skills.
People with vision can use AI as a powerful aid; those without judgment just accept everything.
Harvey notes that trained, experienced individuals see AI as a powerful tool because they know what’s “good,” and they can improve AI outputs. But newcomers who don’t know what’s good tend to accept AI’s answers blindly, which is dangerous.
The truly skilled use AI to reach 80 points and then refine to 100. But those without judgment will just settle for 80. Today, 80 points have little value because AI can produce that instantly. The market demands 100 points—no matter how much AI helps, you need to add that final piece. The prerequisite is knowing what’s missing.
The key skills for future workplaces: proactive judgment, proactive correction, and proactive self-upgrading.
Harvey expresses gratitude for guidance and correction along the way. Being criticized helps develop judgment. Now, things that once required human effort can be replaced by AI. So, is there still a need to train newcomers?
He emphasizes a critical judgment: future organizations will increasingly retain those willing to push themselves to 100 points. If you only deliver 80, AI can replace you. As decision impact and scope grow, top decision-makers share a common trait: their judgment is solid.
Personality traits like anger, gentleness, introversion, extroversion don’t matter; what matters is “vision.” Every step taken now is meaningful—including being criticized and corrected. Looking back, these are valuable experiences. But the current problem is: no one is obligated to teach you anymore. He encourages newcomers to actively seek feedback. Instead of waiting passively, now you should proactively ask for critique. Ask: How can I improve? Can I do better?
Everyone is now looking for “someone who can reach 100 points.” So, the key skills for future workplaces are: proactive judgment, proactive correction, and proactive self-upgrading.
Because of this, no one will do it for you anymore.
This article: AI can do 80 points; those who can’t reach 100 will be eliminated! McKinsey and Harvard alumni advise newcomers to do this. Originally published on Chain News ABMedia.
