Employers are already harnessing AI tools like ChatGPT for text, images and other content creation. But the future of AI for employers is constantly evolving, and you’ll probably find yourself having to adapt and reassess how you’re using it. With key insights from AI expert and Indeed's Senior Strategic Advisor Matt Burney, we’ll explore the future of AI in the workplace.
The current AI landscape for employers
More and more employers are using text generating AI tools like ChatGPT for myriad purposes like creating job advertisements, blog posts, interview questions and employer branding. According to a survey by Capterra Australia, one-in-five Australian SMEs are currently using AI tools in HR, with 67% of employees trusting in the use of AI for training and upskilling.
It doesn't stop here: as we found in our previous article on ChatGPT and automation, companies are using AI to organise their unstructured HR and recruitment data. And as we found in our article on what AI can do for your recruitment — and what it can’t, AI may be able to reduce bias in your recruitment screening processes (to an extent). This is helpful to recruiters, who are 'always trying to automate processes' as well as improve their quality.
What about for candidates? From our July 2023 Australian Market Update, there has been a slowdown in economic activity and increase in the cost of living while job postings decreased. As a result, candidates might increasingly turn to AI tools to help them write more convincing applications, particularly if their writing skills or understanding of keyword usage is lacking.
But on the other hand, University of Edinburgh lecturer Shelagh Green told Teaching Matters that there had been a 'decline in applicants making use of ChatGPT', due to a lack of understanding as to how to create good outputs from the inputs that they chose: what is known in information technology as 'garbage in, garbage out'. This might mean that AI and ChatGPT is more useful to employers than candidates at this point in time – they simply know how to use the technology better.
When it comes to the tools currently available at the moment, it can seem daunting to choose the right one. But as Matt Burney points out: ‘It’s kind of the wild west and the gold rush at the moment. That will consolidate down to bigger players. But making sure that the human piece in there is front and centre is the really important consideration for us in recruitment’. So with this point in mind, let’s look at what the future of AI in recruitment might be.
Where is the future of AI headed?
When it comes to the future of AI, employers and candidates alike are excited and concerned for the potential implications of using AI. But where are things currently at, and where can we potentially progress from there? Let’s explore.
AI as intervention, but not a replacement for human decision-making
Firstly, how far can we currently go when it comes to letting AI make key decisions during the recruitment process? According to Matt Burney, we’re at a point where we can use ‘AI as intervention’ but it’s currently not a replacement for human input, thought and observation. As he explains:
‘If you use a language model to look at interviews, take notes and structure those notes to understand what worked and what didn’t work, you can really start to understand quickly who is good at interviewing, who is bad at interviewing, what different people and behaviours look like. And then applying that to the success or failure of the people who they have ultimately hired.’
The obvious endgame I guess is intervention. So to what extent do we allow the machine to tell us whether someone should be hired? Looking at behaviours in application processes, behaviours in interviews. Could we then say that “typically, people who do that x do very very well?"'
He notes that ‘We are a way off that at the moment, but it’s part of the evolution that we’re seeing right now’.
This means that, for the time being, it’s preferable for HR and management to make that decision themselves. Employers shouldn’t necessarily jump the gun and start looking at using AI in this way now. But in the future, it’s entirely possible that AI could become more and more involved in key decision making. This makes it something to consider as a potential outcome of the way that AI is evolving in the recruitment industry.
The democratisation of AI
Another point for employers to consider is whether AI is truly democratised, and whether it will be more so in the future. What does ‘democratisation’ mean in this case, exactly? Well, it very much depends on your definition. As we’ve seen AI can be used – and may increasingly be used – to make better recruitment decisions, and it’s becoming more available as a tool to larger numbers of people.
This means that more candidates will be using text-generating AI for writing applications, and likewise employers will be using them to create interview questions or job advertisements. But does this mean that these tools are becoming increasingly democratised? Matt Burney doesn’t think so:
‘Many individuals struggle to utilise complex algorithms and models without specialised training. That being said, what we could do, if you look at things like LLama, which is a large language model which can exist on a small platform, you could put that on your phone and create your own artificial intelligence. That to me is democratisation.
'The problem that we have with [most open AI] platforms out there, is that they are large models. There’s a lot of money that goes into them, there’s a lot of resources that sits behind them. That means that we have access to them, but don’t own them. So we’re still beholden to them. They could potentially become somewhat like Facebook or Twitter – where they’re considered a public utility, but they’re not necessarily things you’re in control of.’
You might be wondering why this is an important consideration for employers. One such issue – which we looked at in our previous article on AI and ChatGPT, is what happens to the information that you input into these AI tools. At the moment, we don’t know how this information is being used after it is inputted, making it a potential data privacy risk.
In our last post on ChatGPT and automation, we found that employees who are inputting confidential data into ChatGPT and similar AI tools may be running foul of data privacy laws in Australia. Therefore, employers can look at building their own AI to make data-driven decisions without running the risk of their confidential data being harvested by large AI platforms.
Learning to use AI effectively is key to recruiters
While the future of AI has a lot of unknowns, we can predict that the tools available on the market to recruiters will consolidate down to a few major players. However, companies can take initiative and build their own AI tools, which they can use to make better data-driven decisions about recruitment processes. Our final take home: remember that the 'people' element of recruitment should ideally be at the forefront of your strategy regardless of whether AI becomes a part of this process. That way, employers can give a more human, more accurate reflection of what it's like to work for them as an employee.
