Career Planning in the Age of AI

It’s crazy that ChatGPT is already almost three years old. Although some types of AI systems were already ubiquitous in many areas, the release of ChatGPT clearly marks the moment AI became mainstream. Since then, it’s only become increasingly difficult to deny the disruptive changes the technology is bringing to knowledge work. As this greater trend becomes clearer, figuring out what we should do about it seems more urgent.

Executives, investors, and researchers at the biggest AI companies are forecasting changes to the workplace as extreme as the plots of many SciFi horror films. Clearly, they’re incentivized to want everyone to figuratively and literally buy into their narrative, thereby increasing the value of their stock options and investments. With other “expert” perspectives, I’m finding it hard to distinguish people’s predictions on which roles and functions will be replaced by AI from their opinions on what work is valuable. At this stage, AI’s ability to enhance and boost knowledge workers’ productivity with a variety of tasks in nearly all sectors is clear and evident. If and when AI tools will start replacing entire roles, disciplines, and departments, however, is a different question. I don’t believe anyone currently has the answer to it. Personally, I find it hard to imagine a world in which Software Consulting and Development are fully automated by AI. Obviously, my opinion is biased by the work I do and find valuable - but I believe the same is true for everyone else.

Despite clear signals that AI will fundamentally change knowledge work, the details of how it will do so have yet to be revealed or are being lost in the noise of the current hype-cycle. That being the case, the main question I’m wrestling with is ‘what should I do about it?‘. Here are my thoughts.

The Dystopian Base Case

To get the most depressing piece out of the way first, I’ll start with what I’m coining as my ‘dystopian base case’. In this scenario, AI systems fully replace what I currently do in my field and what I might ever be capable of doing. Everyone currently working in software becomes unemployed and can’t even scrape by driving for Uber because of self-driving cars.

Sounds bad, right? Of course! So what might I do to prepare for this scenario? Oversimplifying, I can really only think of one thing; pivot to a trade or human-focused work. Plumbers and electricians are far less at risk of being replaced by LLMs than Developers, Marketers, and VPs of whatever. Hospitality, social work, and mental health also seem likely to always have demand for a human touch. Although, the number of people relying on ChatGPT as an unqualified therapist justifies some doubt about even this point.

With this scenario and a rough plan of action on how to prepare for it, the final remaining question is: do the potential benefits of taking these actions now outweigh the risks of taking those same actions unnecessarily? For me, the answer here is a clear no. I enjoy what I currently do and am, fortunately, pretty well compensated for it. Any career pivot would therefore, at a minimum, have a significant opportunity cost. Should my nightmare scenario not occur, I’d also have given up higher future earnings potential by pivoting to a less well-compensated field.

It seems simple, because it is, but this mental model and line of reasoning, inspired by Tim Ferriss’ fear-setting exercise, has been tremendously helpful for me. Whenever I become unsettled or anxious hearing or thinking about “the AI future”, thinking of the worst scenario possible, distilling it down to its actionable parts, and running a rough risk assessment against them helps me reestablish a sense of calm and agency. Being confident that there’s nothing it makes sense for me to do now helps me move on more quickly.

Focusing on What Won’t Change

Being content with not doing anything drastic like a career pivot is one thing, but how can we be confident that the actions we are taking make sense given how rapidly things are changing? How should we think about what skills it still makes sense to develop? Or what other paths we should take?

The most actionable advice I’ve come across for this dilemma is from Shane Parish. In a blog post reflecting on the decision-making patterns of Warren Buffett and Jeff Bezos, he summarizes it as follows:

“While we can’t predict the future with any degree of accuracy, we can position ourselves to thrive in multiple possible futures. The key is often found in what doesn’t change.”

Bezos and Buffett were never able to predict exactly how technology and industries would change over time. Instead, they succeeded in different domains by focusing on the same thing—things they were confident wouldn’t change. Customers would always want their products cheaper and delivered faster and more conveniently. Companies built on strong, hard-to-replicate, and proven competitive advantages would always find it easier to grow than companies built on temporary trends and tailwinds.

Applying this same logic to the current situation, our question then becomes: what will AI not be able to change? Focusing on those things ensures that regardless of the specifics of how things play out, the career investments will still pay off. What follows are my very personal answers to the question above.

What Will Stay Valuable

Deep Expertise

Regardless of how the specifics of AI transformation play out, I’m confident that rare and valuable skills will continue to be the primary currency of employment. Borrowing Cal Newport’s terminology from his book So Good They Can’t Ignore You, these are hard skills which are hard to acquire and valued by the marketplace. They’re built through focused deliberate practice and experience acquired in the same or related domains over a long period of time.

Clearly, the specifics of what’s rare and valuable can change because of AI, but change in the value of knowledge work has already been a constant for much longer. Instead of prematurely speculating on which specific skills will be more or less valuable, I’m focusing my efforts on deepening my existing expertise. A key part of that process is actively using AI to enhance my capabilities and support my learning goals. I believe that staying curious and flexible, by integrating new tools and building adjacent skills, is what will leave me best positioned for any major disruptions and opportunities in the future.

Professional Certification

In the job market, it’s practically impossible for recruiters to verify the skills and attributes required for a job in a cost-effective manner and with a sufficiently high degree of certainty. Because of this, recruiters rely on signals. In Signaling Theory, a signal is an action or attribute that is easily observable and verifiable by the recruiter and hard for an unqualified candidate to acquire or fake. These attributes enable recruiters to use signals as proxies for the skills and attributes relevant to the job and for identifying qualified candidates.

AI has already completely changed the signaling landscape and seems likely to continue to do so. Cover letters easily written by AI have already completely lost their value as signals. As AI-assisted projects look better, the developer portfolio also seems likely to lose its signaling value. And to top it all off, new AI products such as LeetCode Wizard are even helping candidates to fake technical interview screens.

Irrespective of how the landscape continues to shift, I’m confident the need for and reliance on signals will remain. Because of this, I’m betting more on the increasing value of high-quality professional certifications. Specifically, any certification from a respected authority with a hard, in-person or online proctored exam with strong cheating protections.

Relationships

If rare and valuable skills are the primary currency of employment, personal relationships and networks are our lines of credit. The references, referrals, and recommendations from people can give us access to opportunities we otherwise couldn’t access. As more signals become easier to fake, I expect the value of personal recommendations and referrals to increase in kind.

Out of the three areas I’ve shared above, this is the one I’ve worked on the least. Working in the same company for an extended period remotely hasn’t put me in contact with a lot of new people. I do a decent job keeping in touch with people I connect with, but don’t get as many chances to meet new people I’d enjoy working with. Because of this, I’m doing my best to collaborate on as many of my side projects as I can and am looking for opportunities to connect with more people in person while I’m living in São Paulo.

Closing Thoughts

I can’t predict exactly how my jobs and career will change because of AI, but I can continue to invest in career assets that will leave me better prepared for (almost) any scenario. I’ve fully embraced AI and LLMs as a powerful tool to help me deepen my expertise, learn new skills, and earn trusted credentials. Unlike ChatGPT’s therapy patients, I’m actively keeping it away from my relationships; I still believe in the value of a human touch. Unless the terminators come for us all, I’m confident this will be more than enough for me and most other knowledge workers to continue to thrive.