Artificial intelligence is everywhere, from smartphones and laptops to everyday apps. It's hard to ignore a booming technology, and understanding it is another hurdle that many of you might be facing. From flashy terms like tokenmaxxing, inference, compute, AGI, and other jargon that can be confusing for everyday users. As AI evolves rapidly, new terms will continue to emerge; therefore, here are a few buzzwords you should know to understand the technology shaping the future.

AI buzzwords you must know

AGI: The term AGI stands for Artificial General Intelligence. Many tech leaders, including OpenAI’s Sam Altman, Anthropic’s Dario Amodei, and others, say that we are close to AGI. But what exactly does it mean? 

AGI refers to an AI system that can think, learn, reason, and solve problems across a wide range of tasks, much like a human. These systems could bring capabilities like critical thinking, reasoning, learning from experience, understanding complex concepts, and potentially self-awareness.

Compute: We can refer to compute as the "engine" that powers AI, as it is the processing power required to build and run AI systems. Therefore, more computing power means faster and more AI models. It includes processors, graphics cards (GPUs), specialised AI chips, memory, and data centre infrastructure to process the power.

Tokenmaxxing: This has emerged as a new workplace trend where employees maximise their AI usage to prove their productivity, measured in "tokens.” Due to tokenmaxxing, companies are spending millions of dollars, compelling them to rethink AI usage. Read more here.

Inference: Inference happens when an AI model uses its learning or training to answer a query or complete a task. This means that every time you ask a chatbot like ChatGPT something and receive a response, the AI is performing inference. In simple terms, it's the moment AI puts its knowledge to work.

Context window: A context window refers to the amount of text an AI can remember while generating a response. Think of it as the AI's short-term working memory. It consists of your commands, previous conversation, and its own responses. A larger context window allows the AI to handle longer documents, more detailed instructions, and longer conversations.

Distillation: It is a machine learning technique that is used to fine-tune smaller, cost-efficient AI models using outputs from a more capable model. This helps the smaller model perform efficiently while using less computing power and lower costs.

AI Hallucinations: This is a term used when AI modes generate incorrect, misleading, or completely made-up results, even though they sound confident and convincing. This could happen due to several reasons, but it could be risky when used in areas like medical, financial decisions, or legal guidance.

Multimodal: A multimodal model is a machine learning model that can understand and work with different types of information, such as text, images, audio, and video. A multimodal model can combine information from multiple sources to understand and respond more effectively.