# BIG AI! How Did We Get Here? ## Essay 2 — Understanding AI Course --- When you use ChatGPT, Claude, Grok, or Copilot, you're using one of the most powerful technological forces humanity has ever created. But have you ever wondered who's behind these tools, how they work, and why they sometimes seem so impressive — and so wrong? In this essay, we'll explore the world of **big corporate AI**, what it can do, and how to use it safely and effectively. --- ## The Big Players — Who's Who When we talk about **corporate AI** or **big AI**, we mean the large language models (LLMs) built by major technology companies and made available through subscription services. Here's a quick who's who: - **ChatGPT** — made by OpenAI ( backed by Microsoft). One of the most well-known AI assistants, available in both free and paid versions. - **Claude** — made by Anthropic, a company founded by former OpenAI researchers. Known for being thoughtful and careful in its responses. - **Grok** — made by xAI (Elon Musk's company). Known for having a more "rebellious" personality and access to real-time information. - **Copilot** — Microsoft's AI, built into Windows and Office. Designed to help with writing, research, and productivity tasks. These are all **cloud-based** AIs. That means when you type a question, it's sent over the internet to the company's servers, processed, and the answer is sent back to you. Your data travels, and those companies store and use it in ways their policies describe. --- ## Corporate AI vs. Local AI There's another way to use AI: running it on your own computer, locally, with no internet required. We'll cover this fully in Essay 3 (Domestic AI), but it's worth explaining the basic difference here. | | Corporate AI | Local AI | |---|-------------|----------| | **Where it runs** | On the company's servers | On your own device | | **Access** | Via the internet | No internet needed | | **Cost** | Free or subscription | Free (software) + cost of computer | | **Privacy** | Your data is stored by the company | Your data stays with you | | **Capability** | Generally very powerful | Varies — often less powerful than big corporate models | | **Internet access** | Yes (usually) | No | The key point: **corporate AI is convenient and powerful, but it comes with privacy trade-offs.** Local AI is more private, but requires more technical know-how and may be less capable. --- ## Getting the Best Responses — A Practical Guide One of the most useful skills you can develop is knowing how to **talk to AI effectively**. Here are some principles that really help: ### Be specific Vague questions get vague answers. Instead of asking *"Tell me about climate change,"* try *"Explain the main causes of climate change in simple terms, suitable for a 15-year-old."* ### Give context If you ask AI to help with something personal or specific, give it the background. *"I'm writing a CV for a teaching job. Can you help me describe my experience?"* is better than just *"Help me write a CV."* ### Break big tasks into smaller steps Don't ask AI to "write me a whole business plan" in one go. Ask for an outline first, then ask it to expand each section. You'll get better results. ### Ask for explanations, not just answers If an AI gives you an answer you don't quite understand, ask it to explain differently. Say: *"Can you explain that in a different way?"* or *"Give me an example."* ### Check and question what you get AI can be confidently wrong. Always think critically about what it produces. Does the information seem accurate? Does it match what you know or what other sources say? **Never accept an AI's output uncritically — especially on important matters.** --- ## Hallucinations — When AI Makes Things Up This is one of the most important things to understand about large language models: **they can hallucinate.** "Hallucination" in AI terms means the model generates information that sounds plausible but is actually incorrect or made up. It might give you a fake quote from a real person, invent a scientific study that never happened, or describe a law that doesn't exist. Why does this happen? Because LLMs are designed to predict what sounds like the right kind of response — they don't have a database of verified facts they can check against. They generate text that *sounds* correct, even when it isn't. **This is not a flaw that will be "fixed" completely.** It may improve over time, but it's inherent to how these systems work. So: - **Always verify important information** from reliable sources - **Don't use AI as your only source** for medical, legal, or financial advice - **Ask AI to cite sources**, then check those sources - **If something sounds suspicious, trust your gut** and look it up independently --- ## Keeping Children and Vulnerable People Safe AI tools can be wonderful — but they also require responsible use, especially when children or vulnerable individuals are involved. ### Things to be aware of: - **Age-inappropriate content**: AI can sometimes generate content that isn't suitable for younger users. It's important to supervise children's AI use and use parental controls where available. - **Data privacy**: Children's information is particularly sensitive. Be careful about what you share with AI systems. - **Manipulation and influence**: AI can be very persuasive. A vulnerable person might be taken advantage of if they're not aware they're talking to an AI. - **Dependency**: While AI can be helpful, it's important that children and vulnerable people develop their own thinking skills and don't become overly reliant on AI for answers. ### Creating an AI Safeguarding Policy If you're an organisation or educator, having a clear safeguarding policy for AI use is essential. Here's a simple framework: 1. **Identify the risks**: What could go wrong when AI is used in your setting? 2. **Set clear rules**: Who can use AI? How? With what supervision? 3. **Train users**: Make sure everyone understands the risks and how to use AI responsibly. 4. **Monitor and review**: Check in on how AI is being used and update your policy as the technology changes. 5. **Have a reporting process**: Make sure there's a clear way to report problems or concerns. --- ## Privacy — What Are You Sharing? When you use a corporate AI, you're trusting that company with your data. Read the privacy policies (yes, they're long and boring, but they matter) and understand: - What data is collected - How it's stored - Whether it's used to train future models - Whether you can delete your data For sensitive work — particularly in education, healthcare, or legal contexts — this matters enormously. **Local AI** (Essay 3) may be a better choice when privacy is paramount. --- ## What Have We Learned? - Corporate AI (ChatGPT, Claude, Grok, Copilot) is powerful and convenient, but comes with privacy trade-offs - Being specific, giving context, and asking follow-up questions gets you much better results - AI can "hallucinate" — generate confident but false information — so always verify important claims - Keeping children and vulnerable people safe requires supervision, clear rules, and understanding of AI's limitations - Privacy policies matter — know what you're sharing and with whom --- ## Glossary of Terms | Term | Definition | |------|------------| | **Corporate AI (Big AI)** | AI systems run by large technology companies on their own servers, accessed via the internet (e.g., ChatGPT, Claude, Grok, Copilot). | | **Local AI** | AI that runs on a personal device rather than on a company's servers. More private, but may be less powerful. | | **Large Language Model (LLM)** | The underlying technology behind chatbots like ChatGPT — a computer program trained on vast amounts of text to predict and generate human-like language. | | **Cloud-based** | Services that run on the internet rather than on your local device. Data is sent to and processed by remote servers. | | **Hallucination (AI)** | When an AI generates information that sounds plausible but is actually incorrect or made up. | | **Safeguarding** | Policies and practices designed to protect people — especially children and vulnerable adults — from harm. | | **Privacy Policy** | A document explaining how a company collects, uses, and protects your personal data. | | **LLM** | Short for Large Language Model — see that entry. | --- ## Further Thinking - Have you ever caught an AI making something up? What gave it away? - If you were setting up an AI policy for a school, what would be your top three rules? - What's the difference between being "sceptical" of AI and being "afraid" of it? --- *Essay 2 of 8 — Understanding AI Course* *Authors: Bea Groves-McDaniel and SAL-9000* *Licensed under Creative Commons (NC, ND) — Share with attribution, no commercial use.*