Exposure Score

An exposure score is a number that expresses how strongly something is touched by a particular thing. “Exposure” means being exposed to or in contact with something. So what is measured is not whether something turns out well or badly, but only how large the point of contact is. One example: researchers assign professions a value for how many of their tasks an AI could take over. A translator receives a high value here, a roofer a low one. The same idea is also used for companies, stocks, or entire countries.

What such a number reveals about jobs and stocks

Major changes are hard to grasp. The question “How much is AI changing our world of work?” cannot be answered in a single sentence. An exposure score breaks this huge question down into many small, measurable sub-questions. This makes it possible to compare professions, industries, or companies with one another.

On the stock market, the metric is used daily. Investors want to know how strongly a company depends on the AI boom. A chip manufacturer has high exposure to AI, a bakery chain practically none. If sentiment around AI drops, the stocks with high exposure are hit first. Anyone allocating their money wants to avoid exactly this kind of concentration.

A common misunderstanding is important to address here: a high value does not mean doom. It means change, and that can go in either direction. A programmer has high AI exposure – and may benefit the most, because the technology takes a lot of work off their hands.

From a task catalog to a metric

First, you need a list of building blocks. For professions, these are the individual activities: writing texts, making phone calls, repairing devices, checking figures. For Germany and the USA, there are official databases containing hundreds of professions and thousands of tasks.

Then someone evaluates each individual task. Can today’s AI take this over, fully or partially? This assessment comes either from experts or, increasingly, from a language model itself. In the end, one adds up what percentage of a profession’s working time falls on the affected tasks. This share is the score, usually between 0 and 1 or expressed as a percentage.

For companies, the process is similar, only with revenue instead of working time. The question asked is: what share of the business depends on the topic under consideration? If a company earns 70 percent of its revenue from data centers, its AI exposure is high. The score is thus only ever as good as the classification behind it. Anyone who divides up the tasks or revenues differently will get different numbers.

Exposure scores in studies, funds, and headlines

The term is most commonly encountered in studies on the future of work. Headlines like “300 million jobs affected by AI” are almost always based on such calculations. Similar figures are also published by economic research institutes and international organizations. It is always worth taking a look at the fine print: affected does not mean replaced.

In the financial world, the metric appears in fund reports. There, for instance, it states how high a fund’s exposure is to technology stocks or a single region. Banks also use the term when quantifying their risk toward a customer or a country.

A third area is IT security. There, an exposure score measures how many vulnerable points a company network shows to the outside world. The basic principle remains the same in all cases: it’s about the point of contact, not a prediction. Related, but not the same, is the risk score. That additionally attempts to estimate how likely and how severe the damage would actually be.