This week, as we head towards ChatGPT’s fourth party, Barry Phillips examines what we know now that we didn’t in November 2022 and how best to use this knowledge. 

Transcript: 

Hello Humans!

And welcome to the weekly podcast that aims to address a key AI development in around five minutes. My name is Barry Phillips

We’re fast approaching the fourth birthday of something that doesn’t feel like it's only been around for a few years. On the 30th of November 2022, a research company called OpenAI quietly put a chatbot online. They gave it the awkward name of ChatGPT that many for a while struggled to pronounce. Nonetheless, within five days, a million people had signed up to try it.

Cast your mind back. Did you have a wow moment when you first saw ChatGPT answer your question just like there was a human being the other side of the screen. What did you think it meant for HR? A clever toy? A threat to jobs? The end of admin as we know it?

Four years on, we don't have to guess any more. We have evidence.

As we approach ChatGPT's fourth birthday, I want to do two things in the next five minutes. First, five things we know now that we didn't know then. Second, five things HR should do with that knowledge. So, let’s get started.

One: our people got there first. Long before most organisations had an AI use policy, employees were already using AI to draft emails, summarise documents and write job adverts. Often quietly, often on personal accounts. "Shadow AI" wasn't a risk on the horizon. It was the starting point.

We shouldn’t have been surprised. There were lessons from history. The Internet came into the workplace in the late 1990s bottom up. It was driven by employees keen to make use of email and the web to help them do their job faster and better. Employers struggled to keep up and in truth they didn’t for a good few years.

Two: it changes tasks, not whole jobs. At least, not yet. The predictions of mass redundancy in year one didn't happen. What did happen is that the work inside jobs started to shift. And the gains are uneven. Confident, skilled users get a lot more out of these tools than people who were never shown how.

Three: the risks were real, but manageable. We've seen AI invent facts, reflect bias and leak confidential information. But in most cases, a human checking the output and a few clear rules contained the damage. The problem was rarely the tool on its own. It was the tool used without judgement.

Four: regulation arrived, and it's looking straight at HR. The EU AI Act treats AI used in recruiting and managing people as high-risk. Whatever side of the Irish Sea you're on, that sets the bar. Hiring, performance and monitoring are exactly where scrutiny will land.

Five, and this is the big one: the barrier was never the technology. It was people. Trust, skills, fear and change management. And that, colleagues, is HR's home turf.

So what do we do with now with all this knowledge? I think it’s five things.

First, swap bans for guardrails. Banning AI just drives it underground. Write a short, plain-English acceptable-use policy. What's fine, what's not, and who to ask. If it doesn't fit on about a page, people won't read it.

Second, make AI literacy a core skill for everyone. Not just the tech enthusiasts. Put the focus on managers, because they set the tone for their teams.

Third, audit the AI already in your people processes. Recruitment software, performance tools, monitoring systems. For each one, ask two questions: who checks the output, and could we explain a decision to the person it affects?

Fourth, redesign roles on purpose. Map which tasks AI is taking on, then decide deliberately what people do with the time it frees up. Otherwise, that time just disappears into more email.

And fifth, keep humans visibly in the loop. Any decision that affects someone's job, pay or wellbeing should have a named person accountable for it. Not "the system said so." Importantly we know it must be a trained human in the loop someone who knows what they are doing there and why they are there.

Back in 2022, the question everyone asked was: "Will AI replace us?"

Four years on, I think the better question for HR is: "How do we best lead people who work alongside it?"

Thanks for listening as always. Until next week. Bye for now.