AI is in the news, in your devices and Google searches, on your social media, and in our government. Even your doctor might be using it to take notes during your appointments. But what is it, and are we using it correctly based on the type of AI that it is?

AI (artificial intelligence) has many uses spanning various industries. The kind of AI you’ve probably heard people mention most recently is AI assistants, or chatbots, such as ChatGPT, Anthropic’s Claude, and Google’s Gemini. MIT’s Hussein Rahnama states that the LLMs (large language models) that power many chatbots have enabled “state-of-the-art performance in text summarization, language translation, question answering, and conversational AI.” These models respond in a conversational tone, ask questions, and give reasonable-sounding answers. But that doesn’t mean that they are always right, or that they understand what they are “reading.” AI cannot always parse meaning in the way that a human can. 

You may have noticed that Google searches now show AI-generated answers before giving you direct links. This model does cite its sources, but it’s still important to check them to make sure it has correctly summarized the information. For instance, I once searched to see if a specific function was available on my iPhone. The AI’s response sounded great and even gave me step-by-step instructions. The issue? Some steps included things the iPhone is incapable of doing, making the entire response useless. AI models sometimes use disreputable sources because they can’t always determine which sources are reliable and which may contain misinformation. While the Google search AI feature can’t be turned off yet, it can be ignored. Many people are switching to search engines and browsers such as DuckDuckGo to avoid AI-generated search summaries and reduce distractions.

Some AI models can take notes during meetings. They can transcribe a conversation, but they may mishear what was said or even record incorrect information. When using an AI-generated transcript, it’s best to listen to the original recording as well, if possible. It would also be wise to review any AI transcripts your doctor is using to make sure they are accurate.

You’ve probably also heard about generative AI. Cole Stryker and Mark Scappicio at IBM describe it as artificial intelligence “that can create original content such as text, images, video, audio or software code in response to a user’s prompt or request.” Many AI models, like ChatGPT and Midjourney (a model that is solely for producing images) can be used to generate works like the following images:

Notice the number of fingers on Grandma, strange placement of fingers on the little girl, odd counter tops and irregular tiles in the background. And why is there a paint scraper?

Beyond the obvious silliness of this image, notice how the handle of the mug disappears into the dog’s paw. AI image generators predict what an image should look like based on patterns in their training data, which can result in physically impossible details.

Critics have pointed out that these models were trained on books, art, and other intellectual property that is not open source or copyright-free, and numerous lawsuits are underway against different AI companies. This is one reason many artists, musicians, and writers oppose the use of generative AI. 

Early AI image generators, like Midjourney, often failed to remove the original artists' signatures within the works it created. A 2023 essay by artist and writer Molly Crabapple in The Guardian exemplifies the backlash against these practices. “Three illustrators launched a lawsuit against leading image-generation companies Midjourney and Stability AI,” she wrote. “… Their complaint claimed that the two companies ‘violated the rights of millions of artists’ … Such attacks on art only reveal the deep anti-humanism of the tech elite. They are a class that shuns human interaction, with its serendipities, annoyances and joys.”

Although the law is still evolving as AI develops, currently artworks created entirely by AI without meaningful human creative input can’t be copyrighted. A 2025 report by the US Copyright Office stated that art created with just a prompt (the description you give AI of the image you want) is unlikely to qualify for copyright protection as there is no human involved in creating the image.

So what is AI? It isn’t magic, and it isn’t human. It predicts, generates, and summarizes based on patterns it has learned. Sometimes those predictions are remarkably convincing. Sometimes they’re confidently wrong. The better you understand what AI can — and can’t — do, the better equipped you’ll be to decide when to trust it and when to verify its answers.

Ultimately, using AI requires healthy skepticism. Ask questions, check important facts, verify sources, and remember that the most important intelligence in the process is your own. 

If you’re already using AI and aren’t sure what type of model you’re using, you could ask the model itself, or check out this article on Medium, which gives a good summary of which models do what.

Have questions about AI? Write to us at [email protected] and we may answer them in a future article.