Optimism for software engineering

AI is the most significant change in software development probably since the discipline was founded. Leaders at top AI labs love to tout as much, including how all white-collar jobs are at risk from this change.

I want to consider - even predict - a scenario that’s much more optimistic for the humble software engineer. I think the AI job doom is overstated.

To be clear - many people will be displaced, but I think free market economics will hold stronger than the media suggests. My prediction is that what once was a company of 100 people will become 10 companies of 10 people, with AI enabling it.

These companies may not be all doing the same thing, or serving the same customers, but many of them will. Everyone will have to use AI to remain competitive, but ultimately scale and competition will be the way wealth remains distributed and jobs aren’t completely scarce. Even AI itself is seeing competition - open-weight and Chinese models have commoditised, and continue to commoditise, tokens. We are moving from ‘which is best’ to ‘which is right’ rapidly.

Leading on from my note about perfect machines, imperfect demands, I believe that the complexity that arises from AI, and the challenge in a company’s ability to translate market needs into products, are both understated.

This presents interesting and novel challenges for the new era of software and economies around the globe. A 100-person company needs a certain level of sophistication in the software tools they purchase. 10 small companies need less sophistication, but more aggregate scale, which presents a step down and out - moving from vertical to horizontal.

Then there’s the challenge of governance, risk and the ability of those responsible to actually exert control over and measure systems which are increasingly autonomous (if you’re entrepreneurial, you just read that as a new start-up opportunity).

So will AI genuinely push technology and humanity forward, or will it just push us through market cycles faster? I predict it will be a bit of both.

Companies whose products are made redundant or uneconomical by AI are in a desperate state and must pivot or die. For companies with products undisrupted by AI who now have to build it into their processes, or for those already struggling to become digital native - AI represents another painful cost of doing business. Most companies are in this bucket.

Some companies will thrive as AI-native and grow to fill gaps created by the uplift in AI, to fill the holes left by dying companies, or to expand into new areas.

Those who use AI to genuinely push humanity’s knowledge and capability will be few, but they will be important. Here’s hoping there’s enough of them.