This is the first in a series on legal issues in Artificial Intelligence (AI). But before I get to the legal part, I want to set out some of the basic ideas of what I will be talking about.

What is AI?

Nobody really knows, because despite all the hype, AI doesn’t exist yet.

The general idea is that computers will be able to think and act like humans. In other words, AI systems will be able to do things like plan, learn, reason, problem solve, explain and perceive. Robots will be also be able to move and manipulate objects. In some definitions, AI systems may also be able to demonstrate creativity and social intelligence.

Then, what’s all the stuff people are calling AI?

John Sumser calls them Intelligent Tools. They aren’t actually intelligent. They mostly just do what they’re told (unlike humans). The scientific community estimates that an actual intelligence will be developed 10 to 50 years from now. What currently exists is sophisticated math, complex statistics, and analytics.

But Intelligent Tools are very interesting and can do things we haven’t been able to do before. This is because computers can store and quickly process massive amounts of data (also unlike humans).

photo of Heather Bussing on HRExaminer.com in black and white

Heather Bussing, HRExaminer Editorial Advisory Board

Intelligent Tools take lots of data and organize, tag, match, compare, rank, average, and calculate based on models and rules. Some also give predictions and probabilities on what may happen. They can find patterns, sort, and connect things in new ways. And they can do old things faster, much faster.

Like all computer systems, Intelligent Tools are all based on rules and logic. This means they are most useful when you want something to work the same way every time.

This is simplistic, because you can add as many rules and contingencies (if x, do y) as you want. The systems can become extremely complex. They can quickly deal with many variables and contingencies, which is what makes them seem like magic.  They can also incorporate the last thing they ‘learned’ into their rule set, making them seem prescient.

But even with all the complexity, Intelligent Tools are systems based on data and rules. Even when you can no longer tell what the rules are, they are still based on rules.

The data is not the thing.

Data is information. In HR, it usually involves bits of information about people. Data is almost always a fact or something you can count, sometimes both.

Facts include someone’s name, address, title, pay rate, manager; they are information specific to that person (although they don’t have to be unique to that person since lots of people may report to the same manager, have the same pay rate or title, and sometimes even the same name).

Often the data is something you can count or measure. For example the number of months or years someone has worked at a company, the time between promotions, how far they have to drive from home to work, how many people they communicate with, and how often they use the word ‘leverage’ in emails.

The more things you can measure and count, the more things Intelligent Tools can do to compare, match, correlate, map, and produce more bits of information about all the other bits. The thing about data is that it makes more data.

But the more you reduce people to bits of information and numbers and create new data about that data, the more you leave out about the actual person or situation.

When you reduce people to specific facts and measurements, then apply rules to those incomplete bits, you always leave out important information. People are not always logical, don’t always follow rules, and tend to change and develop rather dramatically over time. Not everything about a person is suitable to measurement and definition.

This doesn’t mean that the data and Intelligent Tools aren’t useful. They are! Intelligent Tools allow us to look at things in new ways, see patterns and connections we could not have seen otherwise, and detect changes and issues more quickly. They will be an part of strategy and decision making in many aspects of our work and lives.

Yet, there are important limitations to always keep in mind. People can never be reduced to data no matter how many things you measure and how many facts you include. So, the data will always be a representation of some particular aspect of the person at a specific point in time.

This means it’s essential to always ask two questions about any data you use about people:

1. What information is missing that we might need to consider?
2. What is changing or has changed that the current data does not account for?

Most systems and organizations don’t track this information. We work with what we measure and that becomes all we see.

Be careful what you measure, for that’s all you will get.

So maybe it’s time to figure out how to keep track of significant changes in our organizations over time, the ones we know affect people working there such as changes in management, reorganizations, mergers, relocations, headcount freezes, and budget fluctuations.

We also need to stay mindful of what we measure, what behaviors we are encouraging and discouraging with the measurements, and what is missing whenever we draw conclusions from those measurements.

Intelligent Tools help ask better questions; they may not ever have simple answers.


Read the Series

  1. Legal Issues in AI: Data Matters
  2. Legal Issues in AI: Bias and Discrimination
  3. Stay tuned, Heather will have another post in this series soon!

Read previous post:
HRExaminer v10.12

Bad HR data is costing you money, and with automation and intelligent tools, the importance of data quality only increases....