daBongo LMS AI Training Courses

Lesson Content

What Gemini is: the plain-language version.

Google Gemini is a large language model (LLM) – an AI system trained on enormous amounts of text to understand and generate human language. It was built by Google DeepMind, the AI research division of Google. Gemini is one of a family of AI models; different versions (like Gemini Flash and Gemini Pro) offer different balances of speed and capability, and Gemini Advanced is a premium tier offering access to the most capable models.

At its core, Gemini does one thing: it predicts what text should come next given what you wrote. This sounds simple, but the result of doing this at massive scale – trained on vast amounts of human knowledge – is a system that can reason, explain, write, analyze, translate, summarize, and converse in ways that feel remarkably intelligent.

What Gemini is NOT.

Understanding what Gemini is not is as important as understanding what it is:

Not a search engine: A search engine indexes web pages and returns links. Gemini reads your input and generates a response. Even when Gemini accesses the web (covered in Lesson 3), it synthesizes and responds – it does not hand you a list of links.

Not a database: Gemini does not look up answers from a fixed database of facts. It generates responses based on patterns learned during training, supplemented by web access when available. This is why it can answer questions in creative and contextual ways – and also why it can occasionally produce confident-sounding errors.

Not a human expert: Gemini can speak knowledgeably about law, medicine, finance, and engineering. This can feel like consulting an expert. It is not. Gemini has no professional license, no liability, no judgment about your specific situation that comes from years of lived professional practice. Use it as a knowledgeable starting point – not as a substitute for licensed professional advice in high-stakes domains.

Not a factual oracle: Even with web access, Gemini can produce incorrect information – sometimes with the same confident tone as correct information. Understanding this is foundational to using it responsibly.

Gemini's relationship to Google.

Gemini is built by Google (specifically Google DeepMind) and is deeply integrated into Google's product ecosystem. This means:

• Gemini can connect to your Gmail, Google Drive, Google Docs, and other Google services (through Extensions – covered in the features course)
• Gemini powers AI features in Google Search (AI Overviews)
• Gemini is available as a standalone product at gemini.google.com and as a mobile app
• Gemini for Google Workspace brings AI assistance directly into Google's productivity tools

This ecosystem integration is a significant capability that distinguishes Gemini from AI tools with no such native connections.

How Gemini was built: training, not programming.

Traditional software is programmed with explicit rules: if X, do Y. Gemini was not built this way. It was trained – exposed to enormous amounts of text, learning patterns of language, reasoning, and knowledge from that exposure. This has important implications:

• Gemini's knowledge comes from its training data, which has a cutoff date – though its web access supplements this with current information
• Gemini can make errors in areas where its training data was sparse, incorrect, or biased
• Gemini's performance varies by topic – it tends to be stronger in well-documented areas and weaker in niche or highly technical domains
• Gemini cannot be "updated" on a fact you share with it – telling Gemini something new in a conversation does not permanently change its knowledge