Can artificial intelligence and HR actually work together?
When Google’s Chief Decision Scientist, Cassie Kozyrkov, took the stage at UNLEASH London 2019, I knew we were in for a treat. And the presentation she delivered, on mastering business intelligence, did not disappoint. In fact, it got me thinking about the future of artificial intelligence and HR.
You may already know that I am a keen advocate of new technology that helps us work smarter. You will also know that I do not believe robots will take over the workforce.
Cassie’s presentation at UNLEASH London helped me further articulate some of my thoughts in this area. In particular, how artificial intelligence and HR will influence each other, in the years to come – and how we must be careful not to see AI as a lazy solution to our human shortcomings.
Let’s explore in more detail.
AI is just another tool
When I use the phrase “artificial intelligence and HR working together”, perhaps it conjures up an image of two different entities, a robot and a human, each with different skills, different abilities, and different thoughts and feelings, working together. But, spoiler alert, it turns out that we don’t actually “work together with robots”. Well, at least no more than we “work together with hammers”.
People often talk about AI and robots, as if they are living, conscious beings. But according to Kozyrkov, although it might feel like robots are scarily close to becoming our equals – or even our superiors – artificial intelligence is still just another tool.
“Tools we choose to use are always better than humans at something” she explains. “We use them to let ourselves do more. AI is just another tool, and the way you use it is up to you.”
Kozyrkov uses the example that we don’t work together witha hammer. We use a hammer to help us do a job. Because, let’s face it, hitting nails in with bare fists isn’t a particularly fun prospect.
And AI is no different. Just like a hammer, AI is better at doing things for us, such as crunching lots of data in a short space of time. Just like a hammer, AI is just another tool we have invented, to help us do a job we are not very good at.
The skill of the decision maker determines the results
But while AI is good for crunching data, finding patterns, and predicting future behaviours, it is still very much limited by the skill of the decision maker. And if you’re wondering who the decision maker is in this context, it is the person who programmed the AI.
“Unreliable workers can be better workers, if they have bad decision makers as managers” says Kozyrkov. “This is because when they are given bad instructions, they don’t execute them reliably! Put a reliable worker under a bad manager, and you’ll get a reliable execution of bad decisions. And if you put AI in the hands of a bad instruction giver, you will get the most reliable worker of all – and you will have a super reliable, scaled-up version of those bad decisions!”
In other words, while AI is very good at executing commands, you have to be certain that the commands being given, are good. Otherwise, AI can end up actually hindering your success.
So what does this mean for HR?
There are many ways HR and AI can work together
Popular thinking states that there are many benefits for combining artificial intelligence and HR. After all, machines can crunch data in ways humans cannot. For example:
Removing bias from the recruitment process
Predicting an employee’s ability to meet key performance indicators
Executing repetitive HR admin tasks with no human error
But while this all sounds very positive, it’s not always as clear cut as it sounds. Why? Because not only is this all influenced by the way the AI has been programmed, but it is also influenced by the people in charge of collecting and producing the data the AI is processing.
Data is still just a biased text book
“Why do we give Data a capital D!?” Kozyrkov asked the room at UNLEASH London. “Data is just another biased text book, written by humans. And if you want to reach the right conclusion, you need to have different text books, written by different groups, with different biases.”
Simply using AI to process data will not do magical things. You won’t instantly remove bias from the recruitment process, simply by using AI. You have to feed the right datasets into the right machines.
“You need diversity of data” says Kozyrkov. “Even with big data, diversity is still super important. If you’re not careful, you will end up with nothing more than an echo chamber, replicating the thoughts and biases of the people who put the dataset together.”
AI won’t solve HR’s problems, but you can still use it as a tool
Just because we now have the programming skills, the computing power, and the quantities of data to use machine learning, doesn’t mean that we are suddenly going to solve all of our HR issues. Our datasets are still full of bias that we can’t even see, and our robots are still restricted by the people giving them their instructions.
However, none of this means that AI and HR cannot work together. Or, rather, none of this means that AI cannot be a useful tool for HR. And indeed there are many examples of where HR is moving from gut-based decision making, to data-based decision making.
But if you blindly trust AI and big data to solve your HR issues, then you’re probably setting yourself up for disappointment…
Or maybe you’re not. After all, if you create an echo chamber of bias, then the results you’re getting will feel right, because they align with your own beliefs. And you’re therefore probably not going to notice that there is anything wrong.
Discover all DBS Partner solutions for complete human resource outsourcing, including payroll in the Brazilian market at https://www.dbspartner.com.br/dbspartner-english.