By KIM BELLARD
The most (unintentionally) amusing story I read this week was Tim Higgin’s Wall Street Journal article Alex Karp Is Saying What Every Angry CEO Is Thinking About AI. Dr. Karp (yes, he has a Ph.D.), co-founder and CEO of Palantir Technologies, is upset about how AI companies are using relationships with their business customers to harvest data and business insights from those customers. “Something has gone completely wrong,” he fumed.
Now, this is Palantir, mind you; it may not have invented surveillance capitalism but it might have perfected it. It has become essential to government and large corporations across the world. Most of us are aware of how tech companies like Meta or Google give us “free” services that exist primarily to collect more data on us, which they then use to target ads to us, but Palantir’s data collection and analysis operate at a level we often don’t recognize. But make no mistake; it is using our data, and not necessarily in our best interests.
Mr. Higgins quotes former White House AI czar David Sacks in support of Dr. Karp’s concerns:
Anthropic has launched Claude Science, Claude Security, Claude Legal, and of course Claude Code—each expanding into categories previously served by companies building on top of their models. The pattern is consistent: Watch where value is being created, then move in directly. Dominate the model layer, then use that position to capture the most lucrative verticals.
So it is delicious irony that Dr. Karp and others are finding themselves at the wrong end of the power inequality with their data.
I find myself thinking about healthcare when I think above this new wave of data collectors/ synthesizers. It seems pretty clear that the AI companies aren’t going anywhere, and are expected to reshape most industries, including healthcare. Lots has been written about AI’s use in healthcare, including by me. It is both inevitable and, in many cases, desirable. Now this issue of AI’s insatiable appetite for data makes me wonder if we’re looking at things wrong.
I’ve worked in healthcare for longer than I care to admit, and at no point did people not complain that healthcare in general, and health insurance in particular, was too expensive. And yet, costs have kept rising. We’re closing in on $6 trillion in U.S. healthcare expenditures. No matter what kind of health insurance you have – large employer, small employer, ACA Marketplace, Medicare Advantage, even Medicare Supplements for traditional Medicare – your premiums (and/or out-of-pocket costs) are likely going up at rates we haven’t seen in years.
Two well known facts about rising costs are, one, that it is not so much we’re using too many services as it is that Americans pay way higher prices for healthcare than in most countries, and, two, that a relatively small percentage of people account for the vast majority of healthcare spending. The latter has an insidious effect on health insurance premiums, as people with fewer expenses are less likely to have or keep health insurance, making premiums for the remaining people higher. Nobody wants to pay for the people who use a lot of health care, but they want other people to help pay if they end up being one of those people. It’s a conundrum.
Now, optimists hope that AI can do a better job of identifying all the wasted, unnecessary, or inappropriate care we use – estimated as much as one-third – and help make administration more efficient; current levels are estimated as 15-30% of spending. Good goals, both of them, and it is entirely plausible that AI can help with both. But it would still remain that sick people are the “problem” with our health care spending and health insurance premiums, and I want to propose a different way of looking at them.
Healthcare generates massive amounts of data, increasing all the time. Some estimates put it well in the exabyte level, which, trust me, is way more than any of us can comprehend. We generate data when we go to the doctor, when we get lab work, when we fill a prescription, when we go to the hospital, even when we use a wearable like a smartwatch. All those health insurance claims and all those healthcare bills generate data. And, yet, most of that data isn’t effectively used, which I sure hope AI does something about.
So we have a system in which the people who use more health services generate more data, and an AI industry that craves data. This seems like it should be a match made in heaven.
Why couldn’t we have a healthcare system in which AI companies pay people generating healthcare data for that data? I.e., instead of heavy users of healthcare being drivers of spending, they become a valuable resource? And, oh-by-the-way, why aren’t we being paid for our data?
Our data, healthcare data included, is being shared, bought and sold now. Sometimes it is deidentified (supposedly), sometimes not. Either way, we’re not the ones getting paid for it. That should change.
Now, realists will point that that “value” of our healthcare data is nowhere near the costs of our health care, so paying for the latter with the former is impractical. I’ll grant that is currently true, but I’ll also ask: why is that?
I’d argue that our health data is grossly undervalued, because the companies using it are used to getting it so cheaply, and that our health services are wildly overpriced. Reorienting the system so that the former funds the latter should bring them closer into equilibrium.
If data is, as has been said, the new oil, then I’ll point out that oil was also once very cheap, until enterprising people figured out that they could control the supply and thus raise the price virtually at will. We should be those people when it comes to our data, especially our healthcare data.
So I’ll be amused at Dr. Karp being faux outraged at other data companies profiting off of his company’s data, and I’ll hope that we have a fundamental rethinking about who generates value in our data world and how that value is realized. There can’t be a better place to do this than in healthcare.
Kim is a former emarketing exec at a major Blues plan, editor of the late & lamented Tincture.io, and now regular THCB contributor
