IT Home On November 2, it is said that artificial intelligence will not disappear. How should we coexist with it? To be more precise: In the era of artificial intelligence, what will we use to make a living?
There are basically two common answers: “learn to use artificial intelligence,” or find a profession that cannot be replaced by AI and will not be automated, preferably a hands-on job.
According to Business Insider, Ethan Mollick does not believe this approach is feasible. Mollick is a management professor at Wharton School of Business and a highly influential researcher. He focuses on analyzing what artificial intelligence can do today and what it means for ordinary people who need to work with AI. His new book, Co-Existence, suggests that the iteration speed of large models far exceeds most people’s understanding. The only way to cope with its rapid development is to learn how to coexist with it.
This means admitting that many of the abilities you currently possess will lose their value over time; you need to focus on developing those personal traits that AI cannot replicate.
IT Home noticed that Peter Kaffka discussed AI in relation to work, career choices for the next generation, and what will happen to traditional talent development systems when they collapse under the impact of AI, with Morik in his podcast segment “Channels”. The following is an excerpt edited from the interview.
Peter Käfer: Those who worry about their own or their children’s employment are often advised to become AI experts, or to find jobs that aren’t affected by AI, such as plumbers or nurses. Is this advice useful?
Ethan Morlik: I don’t think it has much impact. Let me say this first: even if you choose such a profession, things will be different when robots like PlumBot 6000 come out someday. I’m not an expert in the robotics field, but the attitude of robot researchers I know has changed. They used to think that “large language models and robots are completely different things, and robot development is extremely difficult,” but now they realize that there is indeed a possibility of integrating them.
I don’t know where the future will go, and I believe no one can predict it accurately. These words come from an AI researcher, which may make people feel at ease, or it may cause anxiety.
The industry with the most rapid changes is programming. We have seen transformations taking place in programming positions, and we are even unable to determine whether the total number of programmer roles will increase or decrease in the future. However, it is certain that the work content for programmers will continue to evolve.
The importance of writing zero-error code is no longer as significant as planning clearly what we want to build. Many programmer’s work experiences have improved compared to before, but some feel lost. These roles themselves are constantly evolving.
I mentioned four traits in the book, which can at least help people build some advantages. The first is deep professional knowledge: deeply mastering one field relies more on professional expertise than AI. The second is a broad range of knowledge: being familiar with many different fields.
I have observed a phenomenon in many companies: programmers start doing design work, designers write code, and market professionals draft business plans. The more cross-field knowledge you have, the better you can use AI.
The third is aesthetic judgment. AI produces a huge amount of content; to avoid low-quality outputs, you need to have judgment and be able to select the good from the bad.
The last point is subjective initiative. I don’t think the simple idea of “learning to use AI” has much meaning anymore. We have tested various AI usage teaching methods, and those old techniques are slowly becoming ineffective.
Prompt engineering is becoming more like project management. Instead of relying on any special spells to control the AI, clearly define the requirements: I want this product, what are the acceptance criteria, how to test it, and phased delivery. Provide me with five versions of a solution for me to choose from.
For middle-aged professionals with many years of work experience, who already possess experience and taste, this is good news. But if you just started your career, in an environment where AI can help you search for information and write documents, how can you develop these skills?
There are two levels of challenges: on one hand, at the individual level; on the other hand, it’s our current education and training system.
At an individual level, humans are naturally accustomed to outsourcing mental work—it is ingrained in our instincts. Technology was created to free us from tedious labor. After the dishwasher was invented, no one wanted to continue washing dishes by hand; nor did anyone want to mix butter by themselves.
My child no longer learns to write cursive English, and I myself can’t use a slide rule.
But it is difficult for us to distinguish which abilities can be left to tools. Students cheating on homework with AI is essentially an example of this issue.
Let’s talk about various training institutions. Career learning paths are divided into two categories: high school and university education. This system is indeed a mess at the moment, but there is still room for adjustment. AI is a very excellent tutoring tool, and we can also assess students without relying on AI: written exams with closed books, and face-to-face interviews. In the future, education models could involve AI tutoring after class, while active learning takes place in the classroom, which would actually improve things overall.
I'm really worried about another path to growth: apprenticeship.
In the past, new employees entered banks, consulting firms, auditing firms, and car dealers, following a growth model that has been passed down for thousands of years.
They do dull and trivial basic work with meager salaries, and sometimes have to endure harsh conditions; the exchange is learning skills and building connections, and gaining experience through actual work. Even middle-level managers help with various tasks. Even if the mentor is not good at teaching, new employees can still become familiar with business processes; achieving results leads to promotion, and a continuous stream of qualified workers is produced. This system has been operating for thousands of years.
Now, this system has failed to function properly, and the incentives for all parties no longer match. As a newcomer, the simplest choice is to hand everything over to AI; the quality of output produced by AI is often higher than that of oneself. However, newcomers easily misuse AI: they simply send their requests to Claude and submit the results as they are. They haven’t yet developed judgment skills, and they can’t distinguish whether the report is good or bad.
The manager will also think: since AI can solve problems faster and better, why do I need to hire interns? The corporate management, on the other hand, will consider that there’s no point in slowing down or accepting outputs with unstable quality from interns.
The originally healthy talent cultivation chain was completely broken.
There are only two directions: either develop more formal vocational education programs, especially for basic clerical positions, which will become as important as welding or pipefitting in the future, and follow the path of vocational technical schools; or companies must take training seriously as a serious job. Training new employees requires time investment and short-term efficiency losses, but it ensures a continuous supply of backup talent.
The situation I’m most worried about is when a company chooses to “recruit only experienced experts”. This is a dangerous path: the talent pipeline is broken, and there will be no one to follow in the future.