Transcript:

Hello Humans

And welcome to the podcast that aims to summarise in five minutes or less each week an important AI development relevant to the world of HR. My name is Barry Phillips.

I'm going to keep this one short and practical, because this is a question HR leaders keep firing in to my inbox: which LLM should my organisation actually be using — and is that my decision to care about?

It absolutely is. And with the recent release of ChatGPT 5.4, which many in the industry are now calling the clear frontrunner after some genuinely impressive result improvements from both Claude and Gemini Pro, the landscape feels more confusing than ever. More options. More vendor pitches. More pressure from the executive team to have an answer.

So let's cut to the chase.

First, let's talk about what you're actually choosing between.

Right now, the main contenders for workplace use are ChatGPT , currently in its 5.4 iteration, Claude from Anthropic, Microsoft Copilot and Google's Gemini Pro. Each has genuine strengths. ChatGPT 5.4 has earned its current reputation, particularly around reasoning and handling complex, multi-part requests. Claude continues to excel at nuanced, tone-sensitive writing , which, if you think about the volume of employee communications, policy documents and sensitive correspondence that flows through HR, is not a small thing. Gemini Pro integrates deeply with Google Workspace, which is attractive if your team lives in Docs and Slides.

So here's the first consideration: fit, not ranking.

The "best" LLM in a benchmark is not automatically the best LLM for your HR function. Ask yourself what are your people actually doing with it day to day? Drafting job descriptions? Writing manager guidance notes? Summarising exit interview themes? Pulling together onboarding materials? The answer shapes the choice far more than any industry league table.

Second consideration: data privacy and compliance and for HR, this one carries extra weight.

You are handling some of the most sensitive data in any organisation. Performance records, disciplinary cases, health adjustments, salary information. Before committing to any platform, you need to understand precisely where that data goes, whether it's used for model training, and whether the enterprise tier gives you the controls your legal, IT, and data protection teams require. All the major players now offer enterprise agreements, but the terms differ meaningfully. If you're in a regulated sector, or operating across multiple jurisdictions with varying employment law, you cannot treat this as a box-ticking exercise. Read the terms. Loop in your DPO early. Everyone in the public sector is going for Microsoft Copilot but that doesn’t mean everyone else has to as well.

Third: integration.

An LLM sitting in a separate browser tab that your HR team has to manually switch to will get used inconsistently and abandoned quietly. The ones that embed into your existing tools your HRIS, your document management system, your communication platforms get used properly and at scale. Think about where your HR work actually happens, and favour the tool that meets your people there.

Fourth: cost and scalability.

Per-seat pricing looks manageable for an HR pilot. It can look very different when you're rolling out to managers across the business, because ultimately, HR-led AI tools often need to reach beyond the HR team itself into people managers, recruiters and department leads. Model the real cost at full deployment before you commit, not just at proof-of-concept stage.

Now, and this is what I really want HR leaders to internalise, pick one and go deep.

There is a trap organisations fall into constantly: they run three LLMs at once, nobody becomes proficient in any of them, the outputs stay mediocre, and the whole thing gets written off as "not that useful for HR." That is a self-inflicted wound and it's one HR is actually well-placed to prevent.

Because building capability, embedding good practice, and bringing people on a change journey? That's your craft. Apply it here. Choose a primary platform, build internal knowledge around it, train your HR team and your people managers in how to prompt it well, and treat AI fluency as a professional skill to develop not a vending machine to occasionally prod.

ChatGPT 5.4 may well be the right starting point for many private sector teams right now. But the brand matters far less than the commitment.

Here's what this moment means for HR specifically. The function that has always been responsible for building organisational capability now has to lead by example on the most significant capability shift of our working lifetime. The organisations that get this right won't be the ones that chose the cleverest tool. They'll be the ones where HR built the culture, the confidence, and the competency to use AI well, at every level. That is not an IT project. That is a people strategy. And that is exactly where you come in.

Thanks for listening. Until next week, bye for now.