AI keeps coming up. These are the terms behind it, explained the way a patient friend would. No jargon. No assumed knowledge.
The field asking which uses of this technology are fair, and who gets harmed when it goes wrong. Questions about surveillance, bias, and accountability sit here. There is no settled answer yet, which is part of why it matters.
Learn more in the Plainly app →A set of steps a computer follows to get something done. Your email sorting junk mail is running one. So is the route your GPS picks. The word sounds technical, but it means little more than a recipe written for a machine.
Software that handles work once done only by a person. Answering a question, recognizing a face, translating a menu. There is no robot involved and nothing magical about it. It is a very good program doing a very specific job.
Getting a machine to do a repeated task without someone driving it each time. Automation has been around since the assembly line. AI simply widens the range of tasks that qualify.
AI learned from human writing, so it inherited our assumptions along with our knowledge. If the text it trained on underrepresented certain people, its answers will too. The bias was never invented by the machine. It was absorbed.
Learn more in the Plainly app →Any program you talk to in ordinary sentences. The one on your bank's website follows a script. ChatGPT does not. Same label, very different capability, which is a common source of confusion.
The tool from a company called OpenAI that made most people notice AI existed. You type a question, it types back. Released in late 2022, and the reason this word appears in your newspaper every week.
An AI assistant built by Anthropic, a company started by former OpenAI staff. It tends to be more cautious about the requests it accepts. Useful for long documents and careful writing.
Someone else's computer, doing work your device cannot. When you ask a smart speaker a question, the question travels to a data centre and the answer travels back. All of it in about a second.
Learn more in the Plainly app →Microsoft's AI assistant, built on the same underlying technology as ChatGPT. It shows up inside Word, Excel, and Windows, which is its real advantage. It is already where you work.
A building full of computers doing the actual thinking when you use AI. They consume enormous amounts of electricity and water for cooling, which is why AI's environmental cost comes up in the news.
Video, images, or audio generated to impersonate a real person. The technology improved faster than most people expected. When something shocking arrives, slow down before you believe it or share it.
Software that matches a face in an image against faces already in its records. Your phone uses it to unlock. Police forces and shops use it too, which is where the debate lives.
Learn more in the Plainly app →Google's AI assistant, previously called Bard. It connects to Google's own services, so it can reach current information more readily than some others.
AI that produces something new rather than sorting what already exists. Text, images, music, video. The older kind recognized and classified. This kind writes and draws.
The AI assistant from xAI, Elon Musk's company. It is deliberately less filtered than its competitors, which some people prefer and others find reckless.
When AI states something false in exactly the tone it uses for the truth. It is not lying, because lying requires knowing better. It produced something shaped like a correct answer. For anything that matters, check one other source.
Learn more in the Plainly app →The machinery behind ChatGPT and its cousins. It absorbed a staggering amount of writing and got extremely good at predicting which words should come next. That prediction is what looks like conversation.
Teaching software by showing it millions of examples instead of writing a rule for every case. Nobody described a cat to it. It worked that out from the pictures.
The trained system itself, as distinct from the app on your screen. ChatGPT is the product. The model underneath is what actually generates the words, and it gets swapped for a newer one periodically.
Getting a computer to understand ordinary human sentences. It is why you no longer have to phrase things like a search engine to be understood.
Learn more in the Plainly app →Layers of simple calculations wired together, loosely inspired by how brain cells connect. The resemblance is real but shallow. Calling it a digital brain oversells it considerably.
Models anyone can download, inspect, and run themselves. Supporters argue transparency makes the technology safer. Critics point out that it also removes every guardrail. Both are right.
The company that built ChatGPT. Founded as a non-profit, later restructured to take investment, and now one of the most valuable businesses in the world. That history explains a fair amount about the industry.
What happens to what you type. On most free tools your conversations may be stored and used to improve future versions. Assume anything you enter could sit on someone else's computer, and decide accordingly.
Learn more in the Plainly app →Whatever you type. A short one gets a generic answer. One that explains your situation, the tone you want, and what to avoid gets something truly useful. This is the whole skill.
The system deciding what Netflix suggests, what appears in your feed, and what a shop puts in front of you. It predicts what will hold your attention, which is not the same as what you would choose on reflection.
An automated service that manages investments using rules and algorithms rather than a person. Fees are lower. It also has no idea about your family, your health, or what keeps you up at night.
The date after which the model knows nothing. Ask about something recent and it may answer confidently with old information. Worth knowing before you rely on anything time-sensitive.
Learn more in the Plainly app →Everything the system read while it was being built. Books, websites, forums, articles. Its capabilities come from this, and so do its gaps. What was missing from the pile is missing from the answers.
The 2017 research breakthrough that made today's AI possible. It let systems weigh which parts of a sentence matter most. Deeply technical, and the reason everything changed when it did.
Siri, Alexa, and Google Assistant. They wait for a wake word, then send what follows to a data centre for processing. Worth reviewing what your device has saved.
Recreating someone's voice from a few seconds of recording. Scammers use it to imitate a relative in trouble. Agree a family code word, and call back on a number you already have.
Learn more in the Plainly app →