Effectiveness vs Efficiency, Pareto AI Adoption

Effectiveness is about doing the right things. Efficiency is about doing things faster and producing more of something. For static and consistent tasks, such as booking doctor appointments, writing documentation and billing codes, it’s all about improving efficiency, because there is nothing new to them. This is where AI work can completely automate human work.

But a big chunk of the clinical workflow is multi-task and multi-person; it’s dynamic and does not just change, but expand based on circumstances. The diagnostic and treatment journeys of two patients with cancer but different medical histories, different ages and different attitudes are never the same. In those cases doing the standard procedures but faster doesn’t make for a better outcome. Instead, the task is in making a judgement call and figuring out the right thing to do. It’s about effectiveness.

One of the points from yesterday was about how the atomic unit of the enterprise is the workflow and not the individual worker. The take away is that instead of trying to make individual doctors more efficient at a given task, the goal should be to make the whole end-to-end process more effective. The role of the AI being to augment the human judgement into making better decisions, not just speeding up individual steps, which just moves the bottleneck somewhere else.

This superior, more effective outcome, as argued by Palantir, is enabled by the so-called alpha, i.e. by the best-in-class solution. The alternative is the current tolerable, but suboptimal outcome, which is powered by the beta, or the good-enough solution. The front-line workers such as doctors are interested in the alpha, in maximising the patient outcome. While the back office, where the IT also sits, are happy to trade-off some of that maximisation if it means getting a de-risked option.

Seen from a different angle, this is simply following the Pareto Principle: you can get 80% of the results with 20% of the effort. Investing the other 80% to get that final 20% and reach the maximum is only worth it if it’s going to give you some relevant advantage. And this is where the question of competitiveness comes into play: are hospitals actually competitive with each other? And if not, why would they care about the alpha?