Nicole Junkermann on What Femtech and AI Can Learn From Each Other

It is tempting to describe this meeting as one field helping the other, with the newer and louder one doing the helping. That is not really how it has gone.

The two subjects grew up separately and at different speeds, and each has already worked out something the other is still arguing about. Nicole Junkermann finds that the more interesting way round, partly because it is closer to what actually happens when fields meet, and partly because it sets expectations at a sensible level.

What femtech understands well is that being wrong about a person is expensive in a way that has nothing to do with money. A field that spent years being told its concerns were marginal has a long memory for what it costs to be described badly, and it has developed habits to match: ask before assuming, say plainly what is not known, and treat the person as the authority on their own experience. Those are not soft considerations. They are the conditions under which anyone shares anything true, and everything downstream depends on them.

What the wider field of artificial intelligence understands well is patterns, and in particular how easy it is to find one that is not there. Long enthusiasm followed by long correction has produced a genuinely useful scepticism: an insistence on knowing what a measure actually measures, on checking whether a result survives outside the conditions that produced it, and on keeping the difference between interesting and reliable in plain view.

Put side by side, the two sets of habits overlap a great deal more than they differ. Both come down to a fairly short list.

• Say what is known, and be equally clear about how well it is known.
• Ask rather than assume, then check that the answer was understood as it was meant.
• Test at the edges rather than in the comfortable middle, and expect the edges to be where it fails.
• Treat a correction as ordinary rather than as an embarrassment to be managed.

None of those is technical. All of them are about conduct, which is why the two fields keep reaching the same place along different roads. The glossary holds the vocabulary still long enough to notice the overlap, and the wellbeing technology pages tend to show it in practice rather than in principle.

There is always a temptation, when two subjects meet, to expect the combination to produce something dramatic. Nicole Junkermann has watched enough fields converge to be doubtful. The dramatic part is usually the part that gets announced; the durable part is usually a set of working habits that nobody bothers to describe until much later. The companion note on artificial intelligence in women's health looks at the same ground from the other side, and the note on why healthtech rewards patience explains why the durable part takes as long as it does.

Nothing here is guidance and nothing here is a claim about what any particular thing achieves. It is an observation about two ways of working that have more in common than either usually admits, and which are both better for the admission.

← Back to the blog