In November, we’ll celebrate ChatGPT’s third anniversary, and I think most people would agree that AI is one of those technologies that simultaneously feels old and new at the same time.
We’re still understanding the overall impact that AI will have on jobs; some estimates say that as many as 60% of jobs will require adaptation because of AI. But at this point in time, however significant AI’s positive impact, I firmly believe that HR needs to remain human-first to stay relevant.
People-First (Ask AI Questions Later)
We’ve had a number of terms for HR over the years: personnel, talent management, people management and so on. They’ve all got one thing in common – it’s a discipline that revolves around people. The transformational potential of AI is enormous, but ultimately, dealing with humans needs a human touch.
This is as true at a strategic level as it is at junior, coal-face level: people tend to have an understanding of both broad (strategic) and narrow (human / individual) context, not to mention relationship building, that AI can’t provide today. It may feel intangible, but it’s the difference between the clear rails of a chatbot and the adaptability and warmth of a real person, or a zoom call during the pandemic versus an in-person meeting.
Now clearly, AI can and is having a vast impact, and taking a human-first approach to AI doesn’t mean that humans always do the work first. Generative AI is great at creating ideas, taking a first pass at content or breaking tasks down into manageable chunks. But this always has to be humanised, not least of all to put the incorrect usage of an em-dash back into a sentence!
And that makes an important point: AI might be a fantastic multi-tool, but today it’s still a tool, even if it has the same potential as the world wide web or social media. This means that we have a professional and ethical responsibility to help our teams learn how to use it properly.
The Age-Old Question of Upskilling
Many employees are (understandably) concerned about AI taking their jobs, and with some reports indicating that approximately 80,000 people globally have been made redundant because of AI this year alone, these concerns are clearly warranted.
We can look at this another way. More highly skilled employees are not only more productive, but also more resistant to AI-related layoffs, partially because of their AI skills and productivity, as well as their ability to guide and mentor other staff in the organisation in the same field. Even in areas where staff aren’t being replaced by AI, staff who know how to use AI effectively are outpacing their colleagues in similar areas who are not supercharged by AI.
Again, this comes back to tool use. Skilled AI users understand AI’s limitations and can therefore use it better. A prompt from an experienced LLM user will provide a better output than one from an unskilled user – not to mention an understanding of its restrictions. For example, at the time of writing, the free version of ChatGPT was only able to provide historical data up until June 2024, so any queries about events this year, for example, would either be hallucinations or ‘well-intentioned’ predictions!
Clearly, most large companies will either have access to paid versions of these tools, or their own, in-house versions, but not all will – and some will have policies restricting them. But even in the most advanced, pro user environments with multi-agentic AI, a highly skilled prompt-builder will still need to double-check content, review their brief and modify their requests in line with a constantly changing, human environment.
However, there are areas where AI’s neutrality and independence has very significant potential indeed.
When Human Instincts Fail
For decades, the HR profession has worked hard to eliminate bias and unfairness in working environments. It seems staggering that we are only celebrating the 50th anniversary of the Sex Discrimination Act this year, and that prior to this, it was not illegal to discriminate on the basis of sex or marital status! We continue to see updates to this legislation – not least of all, the 2010 Equality Act which replaced the 1975 Act – but we still need to fight new forms of bias and prejudice in the workplace.
This is one area where AI can support us greatly. AI can identify and help to remove bias in areas like job descriptions and assessments of candidates, because AI tools are effectively ‘blind’ to the process. However, a skilled user will also know that AI systems do need to be built and checked rigorously to remove bias in the first place, but once built, they don’t have good days and bad days and are not prone to lapses in judgment.
And it is worth noting that this checking must be rigorous. A recent experiment by the team at Anthropic noted that one of their AI systems resorted to blackmailing a team member rather than taking the option of being shut down! We should always remember that AI systems are ultimately built by people, and these people are in turn prone to bias. However, AI is fantastic at being consistent once established, and again, this reinforces the need for skilled AI users who understand its limitations as well as its potential scope.
A Bright, AI-Assisted Future
The early stages of any new technology always represent both a challenge and an opportunity. The ecommerce revolution fundamentally changed how we do business. Social media fundamentally changed how we communicate with each other. AI has the potential to make a significantly greater impact on the workforce, and we are still in the age of experimentation. This experimentation is necessary and very positive. However, sometimes this experimentation results in companies prematurely removing positions, which is often neither positive nor necessary.
One thing I would add is that within the HR space, there is a lot of regulation and protected data. AI is well suited to handle this data to some extent, but the application of such data often needs a human touch. For example, there have already been lawsuits against candidate-screening software providers.
AI does give us the opportunity to redesign and rethink how we operate, modernising HR – but this is nothing new. I would wholeheartedly encourage other talent professionals to experiment and investigate, but as with all technology, not to throw the baby out with the bathwater.
Ultimately, HR needs empathy, awareness of individual differences, sensitivity and well-tuned judgement, which (currently!) is something that only humans can provide. Although we don’t know where we’ll be in the next three years, I remain optimistic that we’ll be able to balance the human and the machine – and that organisations that can get this balance right will have a clear advantage in terms of agility and flexibility – not to mention empathy, judgment and humanity.
