There’s a version of “AI does everything” that sounds efficient until you realise you’ve automated yourself out of the actual thinking.
Here’s where it gets interesting: AI-powered research can surface five years of filing history, flag director connections, calculate risk scores, and map ownership structures in the time it takes a human analyst to open a new browser tab. That part is settled. Machines win on volume and speed.
But then what?
Someone still has to decide whether that flagged connection is a red line or a coincidence. Someone still has to weigh whether a company’s shrinking margins reflect a struggling business or a deliberate investment cycle. Someone still has to walk into a room and read the room.
The mistake isn’t using AI for research. The mistake is confusing data retrieval with judgement. They’re not the same thing, and conflating them is how smart people make avoidable errors.
My take: AI should handle the archaeology — digging up everything that exists. Humans should handle the interpretation — deciding what it means for this deal, this client, this moment.
Where it gets genuinely murky is the middle ground. Pattern recognition, anomaly detection, early warning signals. AI is increasingly good at those too. So the line keeps moving.
Which brings me to the actual question worth debating: is there a category of business judgement that AI genuinely cannot replace — or are we just nostalgic about the parts humans used to own?
Curious what people who actually do this work think. Drop your honest take in the comments.
And if you want to see what AI-powered research actually looks like in practice, start at https://borsch.ai