What does... "LLM" actually mean?

Don't you know it? Not so bad. The acronym stands for "Large Language Model". It is often used when someone wants to prove that they know more about artificial intelligence (AI) than their counterpart. AI is used quite excessively in small talk these days. In the topic ranking, right behind "How are you?" in second place, regardless of the person's school-leaving qualification or job. It is used to bluff, brag and bluff by talking about what AI would already be used for and what implications it would have for our everyday lives. It is then usually said: "Your industry will certainly be massively affected." And they always mean the other person's industry.

The impact on their own industry is skillfully downplayed because AI/AI has already been in use there for a long time. As our industry in particular is known for always sailing close to the wind of buzzwords, the abbreviation LLM is used to distinguish it from the AI/AI babble that has become commonplace. But what are LLMs?

Highly complex and yet child's play

LLMs are the basis of AI/AI. Large, so-called language models that have been trained on huge amounts of data. These are transformers that process entire sequences of queries in parallel. And with hundreds of billions of parameters. In everyday language: If this, then this. In educational language: If the hotplate is hot, don't put your hand on it. This process, also known as deep learning, uses interconnected nodes or neurons in a layered structure that resembles the human brain. Sounds mathematical and sophisticated. And it is.

Why is everyone talking about it anyway? Because nobody wants to be one of the diehards, the left behind or the losers whose jobs will fall victim to artificial intelligence. And because we tend to overestimate the impact of a technology in the short term and underestimate it in the long term. Those who constantly use the acronym are also overestimating. It would perhaps make more sense to talk about the learning progress of our own children and grandchildren in order to understand how learning works and how intelligence is created. Children learn to understand basic things such as grasping, crawling, walking, talking, languages and other things through self-learning and are able to practice unsupervised.

So here's a tip: behave like a child and ask the next person who uses the acronym LLM to explain it to you in detail. "What's that?" - you'll be amazed. Those who simply pontificate or parrot will stammer, and they themselves - as children are - will quickly turn their attention to something else because of their evasive answers. Those who have paid more attention in STEM subjects and are most likely to understand something about it will explain the underlying transformer to you in such detail, consisting of a series of neural networks made up of an encoder and a decoder with a self-observation function, that your mind will also be somewhere else. Just like when you read this sentence. So there is still a lot for us humans to learn.

* Benno Maggi is co-founder and CEO of Partner & Partner. He has been eavesdropping on the industry for over 30 years, discovering words and terms for us that can either be used for small talk, pomposity, excitement, playing Scrabble, or just because.