# What the Hell is AI Anyway? > Source: > Published: 2026-08-25 19:54:26+00:00 # What the Hell is AI Anyway? Five years ago, the term “artificial intelligence” sparked images of super-intelligent machines that were either equal to or greater than humans. Whether they [were terraforming earth as part of the Horizon series](https://horizon.fandom.com/wiki/GAIA_(original)), [wiping out humanity as part of Battlestar Galactica](https://en.wikipedia.org/wiki/Cylons) or [just serving as a comedic foil in the Portal games](https://half-life.fandom.com/wiki/GLaDOS), they were the works of science fiction, not science fact. Today, if you say AI, you’ll more likely conjure up images of chatbots, image generators and similar tools. Though more accurately described as “generative artificial intelligence” the term artificial intelligence has become a catch-all term for all of these tools and much, much more. And that’s the problem. AI has become such a buzzword that it has lost all meaning. Virtually anything that is automated in any capacity is now being called AI. For example, Canva claims that its background remover tool, launched in 2019, [was its first AI product](https://www.canva.com/en_in/newsroom/news/leonardo-ai/?msockid=03f8135944606b5a14f9045f45926aa3&utm_source=chatgpt.com). It says similar things about its [auto-resize tools and tools that allow you to connect to other services](https://www.canva.com/canva-ai/). All of this was available on Canva for years and is available elsewhere without AI branding. If every automation tool is AI, then AI becomes a meaningless term in the age of modern computing. While it’s easy to dismiss this as marketing hype and a problem with no real consequences, if we can’t establish what is and is not AI, then how can we set boundaries for what is and is using AI? The first step to regulating something, either formally or informally, is to define it. Unfortunately, “AI” companies have made that almost completely impossible to do. #### The Ridiculousness of AI Over the past few years, we’ve seen some insane examples of “AI-washing.” In September 2023, [Coca-Cola launched a new flavor that was “co-created with AI”](https://www.coca-colacompany.com/media-center/coca-cola-creations-imagines-year-3000-futuristic-flavor-ai-powered-experience) and promised an “AI-powered experience.” In May 2024, [Cooler Master released an “AI Thermal Paste”](https://www.tomshardware.com/pc-components/thermal-paste/cooler-master-introduces-colored-ai-thermal-paste-cryofuze-5-comes-with-nano-diamond-technology) that supposedly improved PC performance. [We’ve even seen the rise of AI golf clubs](https://www.zdnet.com/article/can-ai-make-you-a-better-golfer-i-took-a-whack-with-this-driver-and-got-a-big-surprise/). While it is easy to laugh at some of the more ridiculous examples of AI washing, governments are cracking down. In March 2024, the SEC charged two investment firms for making misleading statements about their use of AI. [They agreed to pay $400,000 in civil penalties](https://www.sec.gov/newsroom/press-releases/2024-36). Though it’s difficult to get a number of the total penalties issued, it is clear that this is a growing space. But even if we steer away from the ridiculous, the term “AI” still covers a lot of ground. [According to Syracuse University’s School of Information Studies](https://ischool.syracuse.edu/machine-learning-vs-ai/), AI is something of an umbrella term that covers not just machine learning, but also rule-based systems and other elements that “enables machines to perform tasks that typically require human intelligence, like problem-solving and decision-making.” But, even within that document, there is room for debate. The article cites Netflix recommendations as an example of machine learning but not AI. But isn’t recommending a film based on another person’s taste something that normally requires human intelligence? The article also claims that both Siri and Alexa are examples of AI, but they have often been very rule-based, often to comedic effect. Why is asking Siri to turn off my lights or tell me the weather an example of AI, but Netflix recommending me an obscure film isn’t? From the end-user’s perspective, it doesn’t make much sense. Automation didn’t begin in November 2022. [Generative AI didn’t poof into existence with the launch of ChatGPT](https://www.washingtonpost.com/pr/wp/2017/09/01/the-washington-post-leverages-heliograf-to-cover-high-school-football/). Computers have been automating more and more tasks for decades. So how do we draw the line of what is and is not AI? #### Looking at Outputs, Not Inputs Ultimately, the technology is likely a red herring. If you use Grammarly to fix your spelling and grammar errors, few will accuse you of using AI. But if you use it to generate paragraphs of new content or to completely rewrite your work, then you are more likely to get called out. However, Grammarly uses the same backend for both. If your goal is to avoid using an LLM or machine learning system, then Grammarly is more or less a non-starter. That’s because LLMs and machine learning systems have brought benefits and improvements to a wide range of tasks, many that were previously possible with “regular” automation. This brings us back to Canva and the background remover tool. [Background remover tools have been around for decades](https://www.plagiarismtoday.com/2013/08/01/how-to-ensure-your-watermarks-are-secure/). However, I don’t know how Canva’s tool works. It’s entirely possible that it uses AI in some capacity to improve its results. But does that constitute using AI? Or is it just using a better background remover? In a similar vein, autocomplete tools have been around for many years, [but they have been improved and expanded through LLMs](https://www.plagiarismtoday.com/2026/06/02/is-using-ai-autocomplete-cheating/). Is using an autocomplete tool using AI? What if it’s powered by an LLM? What if it isn’t but uses another form of machine learning? Between the rampant AI-washing and the expansive nature of the term AI itself, it has almost become a meaningless term from a technical perspective. Something being AI-powered could mean anything from “It generates wholesale content for the user” to “It does slightly better at catching grammar mistakes.” If the technology is meaningless, then users have to look at the outputs. To that end, there’s no firm agreement on what constitutes using AI. Different people, different spaces and different media types have different opinions. However, one general consensus seems to be about who is making the decisions. If the AI is making the creative decisions, then it is AI-generated. If the human is in control, then it may be AI-assisted in some ways, but it is not seen as AI-generated. But that is by no means a universal answer. Many people disagree with that. Some will say that all AI use is acceptable and others would argue that any intrusion on the creative or research process is unacceptable. But if we can’t define AI, we can’t reach that consensus. That, in turn, is the real problem. #### Bottom Line In their push to join the gold rush, the tech industry has failed to establish a clear, enforceable definition of AI. From a technology standpoint, the term is practically meaningless. Even as researchers attempt to cling to their definitions, the industry has hopelessly muddied the waters and the public is rightfully confused. As someone who has been following AI closely for many years, even I don’t have a good, clear, working definition of what it means to “use AI.” [That’s why I have a lengthy AI disclosure](https://www.plagiarismtoday.com/ai-disclosure/), I’d rather disclose any potential “use” of AI rather than risk being accused of violating someone else’s definition. But that creates a major problem. Other than simply disclosing how everything is made, AI or not, there is no way to be sure that your definition of “I don’t use AI” is the same as your audience’s. This kind of disclosure might work somewhat well here, but it isn’t practical in every space or every situation. And that’s why the vagueness of what is and is not AI is a serious issue. How can we establish norms around the use (or non-use) of AI if we can’t reasonably define it? When “using AI” means two very different things to different people, how can we expect to agree on the rules? It is tempting to say that this was a deliberate act. However, the tech industry has long been guilty of overusing buzzwords. Whether it’s crypto, Web 2.0, or the internet itself, tech companies have always used loosely-defined buzzwords for marketing purposes. The idea isn’t even unique to the post-internet era. [As one Redditor put it](https://www.reddit.com/r/AskReddit/comments/1e3tmg6/comment/ldb3q8g/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button), AI is to the 2020s as “nuclear” was to the 1950s. The only major difference is that, with AI, the technology has quickly permeated every aspect of our lives and there’s a real need to establish some kind of boundaries on its use, especially in spaces like education and research. However, you can’t restrict something that you can’t define. And that is the precise problem we face. #### Want to Reuse or Republish this Content? If you want to feature this article in your site, classroom or elsewhere, just let us know! We usually grant permission within 24 hours.