Coding in the AI Age

Last updated: 13/04/2025

Back

AI has arrived. [1] And it seems it's here to stay. [2]

In its current state, I'm not too worried. It's an augmentation tool at best. Useful, certainly, but not going to replace us. My concern is more so what happens if it continues to improve at an exponential pace.

This isn't guaranteed of course. But is likely based on current trends and expert opinion.

My personal definition of AGI is an extreme one, and I've defined it as when I think mass societal shifts will occur. I find it to be a useful model for me.

My definition of impactful AGI: "An autonomous AI system that can perform a job as well as the median-trained human for 1/10th the cost."

Now there are two assumptions baked into my definition and I'll address both:

1. Perform as well as the median human: This is not to say that humans will not be needed at all. We'll still need humans above the systems to manage, monitor and orchestrate them. Whether we need the same amount of humans in that particular job is a different question.
2. Price aspect: This is probably the more difficult of the two to achieve. Based on my current knowledge, I don't believe any of the major AI companies are profitable. They're all growing and investing. Whether it becomes 90% cheaper than a human, I don't know. But based on the marginal costs of software versus the intense costs required for a human, it's viable. [3]

So if we assume my definition of impactful AGI happens within the next decade or so, that begs the question, what do I do? What do I focus on?

First of all, I can either run towards AI or run away from it. [4] I've chosen to run towards it. I want to speak the language of tomorrow. I think it's a wise choice, but time will tell of course. [5]

How to survive in tomorrow's world

The current AI are generative and based on pre-existing data. They can only infer and synthesise based on what we already know. Therefore, if you can be where there isn't a lot of data, aka the edge cases, you'll be much safer.

I think we need to start playing with and augmenting these tools. I don't think being a Luddite is a viable strategy.

But we need awareness. Awareness to know that these tools are not perfect and they can be confidently wrong. However, to be able to challenge them and know when they're wrong, you need the pre-required knowledge.

Focus on the meta-skills:

• Problem-solving: Capitalism rewards problem solving. That is true today and will be true tomorrow. AI is a tool to add to your toolkit to problem solve. For example, the act of writing code is now a commodity. But writing code itself was never the real source of value. Problem-solving was.
• Judgement: In the age of infinite information, knowing what questions to ask and what problems to solve become all the more important.
• Adaptability: Being able to pivot, learn new things and adapt to new situations is becoming more and more valuable as time goes on.

Other skills that may be worth learning: Computer science, prompt engineering, ethics, people skills

I feel that if I can focus on developing these skills, no matter what happens in the job markets, I will be valuable and therefore do well.

I feel if I focus on developing these meta-skills, I’ll be valuable, regardless of how the job market evolves. The tools will differ. The landscape will shift. But the ability to think clearly, solve real problems, and adapt to new situations won't change. That will remain the source of true value.

Footnotes

• [1] The current version of AI is a very specific version of AI: Generative AI
• [2] There are factors that could, of course, change things, such as supply-side issues, like obtaining the required chips, such as if China invaded Taiwan, or economic changes causing funding to dry up and it becomes too expensive to run.
• [3] This includes the price of labour, such as having to pay £3,000 per month for a human versus £300 per month for an AI system. There are of course the ancillary to humans as well. Having to provide equipment, training, office space and so on which increases the disparity further.
• [4] By running away, I mean looking for employment in sectors where AI is not yet possible to have an impact. That is more so the world of atoms rather than bits. Things like being a nurse, a plumber or a builder.
• [5] I think it's wiser for me personally. It gives me the potential to have more impact and has more upside potential. In a manual trade for example, my downside will be limited, but so will my upside. In software you can write one line of code that could impact millions. That's not true for labour based roles.