Note: This article is written with insights I learned from Li Fan, Chief AI Officer of Circle and formerly of Google, Baidu, and Pinterest. I heard her speak at the Ascend Pinnacle Summit, and her talk struck a chord with me. I asked if I could write an article incorporating her thoughts, and she graciously said yes.
“You won’t be replaced by AI. You will be replaced by the person who knows how to harness AI.” – Li Fan
There’s a great deal of anxiety about AI right now. I saw this firsthand as my son was applying to college. While applications to computer science programs remain at an all-time high, many high schoolers are wrestling with the new reality that GenAI has introduced in just a few short years, right in the middle of their high school careers. AI and machine learning have been part of our world for a long time, but the sudden rise of GenAI among the general public makes its potential feel more real—and more intimidating for those who are unsure what the future holds.
But it’s not just future workers who feel it. GenAI is already here, disrupting work behind the scenes, for better and for worse.
I remember when I worked peripherally on the first PayPal site translation (which, by the way, was from U.S. English to U.K. English). We had armies of manual translators who managed string extraction, translation, integration, and then QA and maintenance. Every new language increased the complexity and introduced new chances for error. Today, AI can substantially reduce the cost of translation while speeding up innovation. It’s not just faster; it completely changes the game. Companies can now go global sooner and reach new markets more easily.
The same is true for customer service: New GenAI tools offer faster, more accurate answers to routine questions, freeing up agents to focus on the most complex problems, but also requiring fewer humans at the other end. We’re even seeing this play out in transportation; Waymo, without human drivers, now does almost the same number of trips in San Francisco than Lyft does with human drivers. (Fun fact: In The Fifth Element, they have flying cars and intergalactic travel—but the taxis still have drivers! Goes to show how hard it is to imagine specific technologies until they’re already here.)
As with any wave of technological change, there are winners and losers in the AI landscape. That’s why future-proofing your career matters now more than ever. Long-term relevance will come from investing in the skills that remain irreplaceable—and learning how to leverage AI itself, no matter your career or industry.
Job evolutions in history
Recently, Bill Gates said that in ten years, humans likely won’t be needed “for most things” thanks to AI. That struck me, because I remember hearing similar stories when I started at BCG in 1998. Certain partners would talk about their mentors, the generation of consultants who made slides by hand using lightboxes. Some still drew them on graph paper and handed them off to executive assistants to convert into client decks. As associates, we were PowerPoint masters. Every generation got better with the tools, but that just raised the bar for the quality of the work, not eliminated it.
My mom is another example. She was a very proficient keypunch operator. Never heard of it? That’s because by now, the job is completely obsolete. Her job was to enter timecards and employee data into an IBM machine big enough to fill half a room. The data was stored on physical punch cards (hence the name), which were moved into boxes and archived for future access.
These machines—and the job itself—were sunsetted in the ’90s. At their peak, tens of thousands of people were employed as keypunch operators. The work was monotonous and tedious, and my mom had to work the late shift so the timecards could be processed before the staff came back the next day.
Now? No one even remembers that the role existed.
Five Ways AI is Changing Work
With all this in mind, plenty of people are wondering what changes AI will bring to the work we do and how we do it. I think it’s worth looking at these in more detail—both the possible effects this technology may have and the effects we’re already seeing right now. Here’s my take:
1. AI will cause some jobs to consolidate and/or become obsolete.
AI is supercharging many industries and making entire job categories unnecessary. This is alarming to a lot of people, but it’s worth remembering that changes like this aren’t completely unprecedented. We don’t mourn the historical loss of candlemakers, horse farriers, or laundresses because what followed transformed society. One major financial institution recently reduced its overnight analyst team after AI began summarizing global news across markets for morning review. Now they can cover more sectors and markets than ever.
Content writers, researchers, and customer service agents may be the first to feel this shift, but it won’t stop there. Still, there are benefits, because where AI really shines is in its ability to automate repeatable tasks. This may allow us to focus more of our energy on higher-value, creative, or strategic work.
2. AI raises the floor, but lowers the ceiling.
A recent MIT study showed that GenAI helps those who used to be the slowest or least productive the most. It raises the floor for lower-performers, but those who were already high performers see much smaller gains.
This performance compression also means that the premium for high performers may shrink. A 10x engineer used to be worth their weight in gold, but with GenAI, more engineers can demonstrate higher levels of productivity as recently posted by Gokul Rajaram. The top will still command a premium, especially in AI, but the gap between “good” and “great” may shrink at most companies.
3. AI will democratize access to knowledge and reduce information scarcity.
Doctors and nurses are one of the most constrained resources in society. Ezra Klein once pointed out that the U.S. doesn’t manage this with rationing like other countries; we manage it with pricing. But what if AI could ease that constraint? Already, people are using AI for quick diagnoses before heading to urgent care. I did so when I had a dizzy spell recently. I took my blood pressure and asked AI to give me advice on whether I should be worried. It gave me quick and easy advice that saved me a trip. This capability could potentially lower the burden on providers by helping people get quick answers they might have otherwise had to seek from a doctor.
Interestingly, AI chat therapists are even rated as nearly indistinguishable from human ones and often get higher ratings for their (written) responses. This break in knowledge bottlenecks will improve lives, but also lower demand in different sectors.
4. AI will exceed people in many areas.
We pride ourselves on our ability to think and adapt—but what happens when that’s no longer a differentiator? What happens when AI does it better, faster, and cheaper? Again, these questions can cause panic, but they aren’t a reason to avoid AI. Rather, we should be doing what Fan suggests: learning to harness it. Being better at understanding, assessing, and creating via AI means we are upleveling our skill sets.
A recent study of data scientists shows how AI can improve productivity. This could mean one of two things: Either we end up with fewer data scientists, or the ROI on them becomes so good, we decide we should have more, because revenues and profits increase thanks to this improved work. The world could go either way, so be the person who learns to harness these tools.
5. AI will redefine which skills are valuable.
The rise of AI in many fields means that human input will shift to oversight, interpretation, and decision-making. The work itself may be done by AI, but knowing how to guide that work will be key. General writing, basic design, and simple marketing tasks are already being taken over by AI. One company I spoke with shared how their designer now generates all of their website images and icons using AI, then does a final pass to polish them. It saved months on their go-to-market timeline and significantly reduced costs. The designer is still designing, but now she manages prompts alongside pixels. Just like slide producers, keypunch operators, and typing pools became obsolete, skills that AI can do better will fade. However, they will be replaced by skills that require judgment, creativity, and human ingenuity. Future-proofing means investing in the skills that will be hard to replicate.
Five Ways AI Won’t Replace Us
On the other side of the coin, there are also things AI can’t do, and likely won’t be able to do for the foreseeable future. Understanding where this technology falls short is how we find our direction in an AI-driven world. Here are the areas where a human touch will continue to be valuable:
1. Authenticity will still be at a premium.
Li Fan mentioned a story about professors reading student essays. They could easily tell who had used AI—not because the writing was bad, but because it lacked creativity and originality.
One of AI’s biggest weaknesses is that it’s a consensus machine. It averages human thinking, which results in a likely acceptable answer based on available knowledge. That’s not bad. In fact, it is quite useful in many cases. But it’s not magical, either. As Fan said, “AI looks good on the surface, but you can feel it is not real.” Why have live speakers when you can watch a video instead? Why read an article when AI can summarize it? Why listen to a podcast when you can just get the tl;dr? We do these things because they provide a sense of authenticity and connection. The more fake things feel, the more reality we want.
2. Real-time thinking still matters.
AI is trained on the past, so while it can extrapolate ideas about the future, it can’t accurately anticipate it. When we react to current events, make decisions in real time, or synthesize new patterns, we’re doing something AI can’t yet replicate. That’s where humans still shine. History is being made every day, and humans are needed to react and respond in real time.
3. Value will be placed not on the answer, but on the question.
I once saw a video of a guy painting in Excel. He created art out of cells and color fill. The magic wasn’t in the tool itself—it was in knowing how to use it creatively. GenAI works the same way. Those who know how to prompt, validate, and push for better answers will lead.
The best leaders don’t know the answers, but they know how to ask the right questions to elicit the right outcomes. The same is true in a world with AI.
4. Storytelling and the human experience can’t be faked.
What makes writing memorable isn’t just the facts. It is the stories, the vulnerability, the moments we’ve actually lived or can see reflected in our own experiences. AI can make up stories, but they are not real. We connect with truth, with imperfection, with what’s messy and human. We connect to who the author is and what they experienced, not just a nameless, faceless story. Reading When Breath Becomes Air, a book written by a doctor facing his final days, is a reminder that the human experience is truly special and worth investing in.
5. Validation and judgment still matter.
Remember the lawyer who submitted a legal brief full of hallucinated cases written by ChatGPT? That story will repeat itself unless we continue to validate AI outputs. In her talk, Li Fan shared another story: a customer asked why their refund had been denied, and the AI chatbot shared the entire internal decision tree, something never meant to be seen externally.
Boundaries matter. Judgment matters. Those need to be defined and reinforced, and that’s something AI itself can’t do without human input.
When my son asked me what he should study in order to thrive in an AI world, I didn’t have a quick answer for him. But after some thought, we came to a decision together: He’s now in college studying data science. Knowing how data works, how to capture it, structure it, and find meaning in the patterns, is already essential. Even more important is learning how to ask the right questions. In a world shaped by AI, curiosity and judgment will still be the most powerful tools we have at our fingertips.
How to Stay Relevant in an AI WorldRead MorePerspectives
