In a recent blog post, CFO advisor and friend of the list Lauren Pearl discussed the impact of AI on the age-old software “build vs buy” decision.

In short, Lauren argued that AI-driven vibe coding has made building software in-house far more appealing when the stakes are low, but when stakes are high, buying is still the way to go.

To quote Lauren:

Her argument can easily be extended to the provision of data science advice.

AI has made it easier than ever for organisations to “build” their own “expert” data advisors. For high-volume, low-stakes situations, this may be the best approach.

It’s essentially the principle that’s always driven machine learning adoption. And now AI has amplified the gains created by ML automation, by being able to code and train those models faster than any human ever could.

But when the stakes are high, data volumes are too low for ML, and there’s real potential for things to go very wrong - as is the case for many strategic decisions - this is where “buying” human expertise still wins out.

To paraphrase Lauren:

When you “buy” a human data scientist, you don’t just buy analysis.

You buy their support in interpreting that analysis, expert knowledge of the best techniques to use, validated results, trust in the outputs, knowledge of when the work needs updating, and someone to identify the problems when models inevitably break.

So, if you’re a data scientist trying to get ahead in an AI-first world, this is what you need to focus on providing - decision support, not just technical outputs.

If not, you might find the build vs buy decision suddenly working against you.

Talk again soon,

Dr Genevieve Hayes