This week Barry Phillips looks at the AI jargon that is becoming common parlance for the HR professional

Transcript: 

Hello Humans!

And welcome to the podcast that aims to address an important AI issue relevant to HR in around five minutes each and every week.

Today we're going back to basics and talking about AI jargon and more particularly AI acronyms. 

You know the moment. Someone from IT says, "We're thinking of using an LLM with RAG, keeping a HITL of course," and you smile, write it down, and pray no one asks a follow-up question. Well, in fives minutes time you'll be the one asking the follow-up questions. Let's decode six of the big AI acronyms.

Number one: LLM — Large Language Model.

Not, as you might hope, "Large Lunch Menu." An LLM is the engine behind tools like ChatGPT and Claude. It's a system trained on enormous amounts of text so it can understand and generate language. Think of it as a colleague who has read practically everything ever written — but has never actually done a day's work. Brilliant with words, needs supervision.

Number two: GPT — Generative Pre-trained Transformer.

Yes, the "GPT" in ChatGPT actually stands for something. "Generative" means it creates new text rather than just retrieving it. "Pre-trained" means it did all its studying before you met it. And "Transformer"? Sadly, nothing to do with robots in disguise — it's the type of architecture underneath. You don't need to remember the details. You just need to know that when someone says GPT, they mean a specific family of these language models.

Number three: NLP — Natural Language Processing.

For years in HR circles, NLP meant Neuro-Linguistic Programming — that thing from leadership courses in the noughties. In the AI world, it means Natural Language Processing: teaching computers to understand human language. It's what lets software scan ten thousand CVs, summarise exit interviews, or spot themes in your engagement survey comments. Same acronym, very different training day.

Number four: RAG — Retrieval-Augmented Generation.

This one sounds like something you'd use to polish the office trophies, but it's genuinely useful. RAG means the AI doesn't just rely on what it memorised during training — it looks things up first, then answers. So instead of guessing your annual leave policy, a RAG-powered chatbot actually retrieves your real policy document and answers from that. Less making things up, more citing sources. Every HR chatbot should have it, frankly.

Number five: HITL — Human In The Loop.

My personal favourite, because the human in the loop is probably you. HITL means a real person reviews or approves what the AI does before it takes effect. So the AI might shortlist candidates, but a human makes the final call. If a vendor ever tells you their hiring tool doesn't need a human in the loop — that's not a feature, that's a lawsuit with a subscription fee.

And number six: AGI — Artificial General Intelligence.

This is the big, slightly sci-fi one. AGI refers to a hypothetical AI that can do pretty much anything a human can do, across the board. It does not exist yet, no matter how confident that keynote speaker sounded. So if anyone claims their onboarding software is "basically AGI," feel free to raise a sceptical eyebrow. You have my permission.

So, to recap: LLM is the well-read engine, GPT is a famous family of them, NLP helps machines understand us, RAG makes them check their facts, HITL keeps you in charge, and AGI is still science fiction.

Next time the tech team starts speaking in capital letters, you can lean back, nod knowingly — and maybe even throw in an acronym of your own.