When it comes to AI, you don’t need to be an expert in the intricacies of LLM or NLP to sound like you know what you’re talking about (don’t worry, they’re explained in detail below).
Think of it this way: while you might not fully understand how the internet works, you’re probably comfortable talking about search engines, your browser of choice, the pros and cons of accepting all cookies and where your IP address is geo-located.
With AI on course to infiltrate every area of our lives, our knowledge will most likely evolve on a comparable basis, assimilated over years of hearing the same terminology, over and over again.
In the meantime, instead of fearing the many acronyms and terms AI has already produced, it’s a good idea to get to grips with the basics so you can hold your own as the technology continues to unfold.
Below we’re detailing some of the most-used phrases and terms that should help you position yourself as the smartest person* in the room in terms of AI.
Generative AI
Generative AI, or gen AI as it is frequently abbreviated to, refers to artificial intelligence that is able to generate text, images or videos.
Generative AI is trained using vast amounts of information gleaned from many different sources, also known as data sets. What sets generative AI apart from traditional AI is its ability to interpret data patterns in a bespoke way.
Chatbot
As the name suggests, a chatbot is essentially a bot that can simulate a conversation. In the past, chatbots had limited capabilities using pre-set responses (as anyone who has ever had a frustrating ‘Live Chat’ customer service experience can attest).
But thanks to generative AI, chatbots now harness the power of natural language processing and machine learning, and are capable of having intuitive conversations. They can even write poetry, short stories or create detailed travel itineraries. Examples include OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude.
Prompt
In order to get the information you’re looking for from an AI chatbot, you have to give it a prompt. For example, if using a chatbot to help you look for a job, you might ask the chatbot ‘show me marketing jobs in Dublin’ or ‘show me jobs that don’t require a college degree but pay over 50k a year’.
You can also have more sophisticated and nuanced conversations based on the prompts you give. For example, asking the chatbot if it can make comparisons, and offer insights.
For example, a prompt could be: ‘can you compare these job examples and tell me 5 reasons why I should pick one over the other’. The more direction you give a chatbot via prompts, the better tailored the information it will be able to give you.
LLM
This acronym stands for large language model and refers to a specific type of generative AI that has been trained using expansive amounts of data. ChatGPT is fueled by LLMs and this is why it is able to answer pretty much any question you ask it.
NLP
Natural language processing, or NLP for short, is a branch of AI that allows computers to understand, generate and manipulate human language.
Some of the simplest but best examples of NLP are Google’s Alexa and Apple’s Siri functions. They operate by translating voice commands into actions, and it’s believed that advances in generative AI will make NLP even more sophisticated and interactive.
Hallucination
AI hallucinations are when LLMs provide incorrect or misleading information and are caused by insufficient training data which results in incorrect data patterns.
Bias
AI bias occurs when unfair or prejudiced information informs outcomes and is the result of biased or incomplete information being used to train the LLM.
There are several different types of bias including data bias (where the data is skewed or incomplete), selection bias (where the data isn’t diverse enough) and historical bias (such as data that reflects discriminatory bias with regards to race or gender).
*Unless that room is filled with engineers, developers and machine learning experts.
You can find out more about how AI is impacting the workplace and how to get ahead here
