CEOs are stoking anxiety about AI replacing our jobs while using it to justify lay offs. This seems to be the latest chess move by those in power, a retaliatory response to the unprecedented COVID era remote work mandates that briefly gave workers the upper hand. The economy cycles and the pendulum swings back. In this era of so-called end-stage capitalism, bosses will look for any excuse to enact lay offs if it helps the stock go up. There’s a virality to it too— easier for Mark to cut thousands in the name of efficiency when Jeff does the same. Putting the fear of god layoffs in Alice and Bob has been a classic play to extract more output from human capital.
So what’s a middle aged tech worker to do?
On the one hand, I feel liberated. I’ve been fortunate to have studied and developed skills that continue to provide value at work, or at least the appearance of value. Over the last 20 years, I’ve worked in medical devices, robotics, wearables, sensors, algorithms, data, machine learning and AI. I’ve seen the technology hype cycle play out for several of these paradigms. I’ve followed a couple start ups that reached billion dollar valuations. I’ve also worked or almost worked for 4 startups that failed (including my own attempt). I was intrigued enough with FOMO to apply at startups that fit my profile— ultrasound and medical diagnostics— both of which had charismatic founders who played it fast and loose with the science. Those startups were Ubeam and Theranos.
Back then, I was fortunate enough to be able to take the safe bet working for the “fruit company”. Since then, my ten-year tenure here has been a slow and steady climb, with a few R&D dead-ends (big swings) and marginal contributions to shipping features (base hits). I’m back up at bat for what I hope will be some RBIs to close out the remaining innings. In that sense, I haven’t gotten aboard a rocket ship— no rapid climb up the ranks, no overnight life-changing windfalls. But it’s been more like sailing aboard a tanker ship. And thanks to slow and steady appreciation and compounding (thanks AAPL) on top of spending within our means, enabled by the various head starts we received from our parents (thanks mom and dad), we are well on the road to financial independence. If the objective function to measuring success is income to stress ratio, I’ve reached a very comfortable local maximum. This is a small scale flywheel — the more liberated I feel, the more agency I have. With this agency, I’m able to work more creatively and effectively, which in turn amplifies my impact. And knowing that I’m moving the needle in turn makes me feel more secure, and thus more free.
But back to AI. Is this a paradigm shift? Has it affected my day to day work? Is it an existential threat? What’s my advice?
1.) Learn to use the tools. Just as programming and scripting scale a knowledge worker’s output via automation and repetition, learn to leverage AI to automate and/or outsource time consuming, low value tasks. Focus your value add on creativity, higher up on the ladder of abstraction. To use a research lab analogy, free yourself from the work of a technician or undergrad, and think more like the senior author or PI. In many ways this is a huge unlock for individual contributors. I don’t have a team of juniors (and I don’t want one), nor a discretionary budget to spin up a contingent labor force. But a few lines of code and some compute budget spend can get me results 80-90% as good as humans with less than 10% of the time and cost. This is work that I would otherwise need to spend time either monotonously doing myself, or socializing and lobbying for resources to get it done for me.
2.) Don’t discard the fundamentals. At the end of the day, the buck stops with you. An LLM can get you 90% of the way towards solving a problem, say writing some boilerplate code for a routine task. But if it doesn’t work, even after repeated prompt engineering, that last 10% is on you. If you don’t know the programming fundamentals, if you don’t know how to craft a good essay, if you don’t know what is factually correct, if you don’t know the outcome you want, no AI will save you. As we leverage AI, we will skip the fundamentals— when we climb up the ladder of abstraction we will inevitably bump up against the limits of our own training. We shouldn’t constrain our productivity by the knowledge we’ve acquired chronologically. But we should use it to shine a light on where we have gaps. And if that means coming back down to learn lower level fundamentals, then we absolutely should, and treat it as a gift that we were able to so quickly identify where we need to focus. If there are any existential threats, it’s not that people are taking short cuts, it’s that they lose the ability to learn how to do it the hard way. Yes, use AI to craft that essay. But if that essay sucks and you don’t know why and you don’t know how to write a coherent sentence, then the AI (and the companies behind it) have used you.
3.) Double down on humanity. This is a potentially tricky one for kids or people struggling with delusions or with their grasp on reality. AI is not your friend and will never be the cure for loneliness nor a replacement for human relationships. But if you are generally of sound mental health, it can help your relationships, both with yourself and people in your life. And that in turn can reduce suffering. What I mean is, consider using LLMs as a mirror, a tool for reflection. In my experience, it can do some of the lifting of a life coach or low-skill therapist, and in many cases that is perfectly adequate. When you have a difficult emotion, particularly when it comes to a relationship conflict, take the time to pause and write it out. Name it, describe it to the best of your ability, be overly verbose, ramble, say controversial, offensive, shitty things. Just let it all out. Submit the prompt if you want. I’d be willing to bet that the majority of the therapeutic benefit comes from just writing it out. If you do submit, take the response with a grain of salt. Trust yourself to know the ultimate truth, and let the response only resonate when it says things you already know, but are lost in the fog of emotions. And let this be a form of practice. As you build the muscle to transfer difficult emotions from your head into text, it becomes easier to do this each time. Let it lower the barrier of entry to engage a real human therapist should you need further help. And ultimately, put the phone down and let it become easier to share your difficult feelings directly with the person who caused them. That is true freedom.