top of page

"AI Is Inevitable" a Marketing Scheme: A Data Scientist's Take

Writer: Madelyn Clarke
Madelyn Clarke
40 minutes ago
9 min read

Red poster with robotic hand and shadow of fingers crossed silhouette, text reads AI Is Inevitable and AI vs AI FOMO over hype.

I'm a data scientist. I build models in R and Python, I've validated market data for an IPO filing, and most of my week is spent separating signal from noise for clients. So when someone tells me their new software is "powered by AI," my first question is always the same: powered how?


Most of the time, the honest answer is a chatbot with a nicer logo.


My take, after years of building models for a living: a lot of what's being sold to businesses as AI is marketing. The real breakthroughs, the ones predicting protein structures and catching cancer earlier, come from specialized models built by scientists on carefully curated data. The tool that writes your LinkedIn posts is not that. Treating the two as the same thing is how companies waste money, publish junk, take on legal risk and, if we're not careful, get a little worse at thinking.


Is 'AI is inevitable' just marketing hype?


Yes. "AI" has become a label companies stick on a product to raise prices, raise money or look current. It's common enough that there's a name for it: AI washing.


Regulators have noticed. In March 2024 the SEC fined two investment advisers, Delphia and Global Predictions, a combined $400,000 for claiming AI capabilities they didn't have. Global Predictions had called itself the "first regulated AI financial advisor." Six months later the FTC launched Operation AI Comply, including a case against DoNotPay, which sold an AI chatbot as a stand-in for a lawyer without ever testing whether it performed like one. FTC Chair Lina Khan said it plainly: "there is no AI exemption from the laws on the books."


Now compare that with the AI that earns the headlines.


Take AlphaFold2, built by Google DeepMind. It predicts a protein's 3D structure from its amino acid sequence, it raised prediction accuracy from about 40% to 90%, and its creators shared the 2024 Nobel Prize in Chemistry.


Or take the MASAI trial in Sweden, which screened more than 100,000 women for breast cancer, half of them with AI support. The AI group had 9% more cancers caught at screening and 12% fewer showing up between screenings, and radiologists had 44% less reading to do (results in The Lancet, January 2026).


Those results came from purpose-built models, trained on carefully verified scientific and medical data, tested in controlled trials and checked by specialists. The writing assistant inside your CRM was trained to predict the next likely word. Both get called AI. They are not the same thing, and a vendor who blurs the line is selling you the reputation of one with the substance of the other.


Marketing Strategist Viewpoint


The tag line, Ai is inevitable, is deploying a classic FOMO-driven marketing tactic. By framing artificial intelligence as an unstoppable force of nature, vendors manufacture artificial urgency and exploit executive anxiety about falling behind competitors, effectively bypassing the normal requirement to prove a product's actual value. Much like the hype cycles around the Metaverse, Web3, and enterprise blockchain, this narrative shifts the burden of proof onto skeptics and disguises aggressive corporate sales strategies as human destiny. Ultimately, calling AI inevitable is a deliberate sales pitch designed to lock in customers and inflate valuations before buyers have a chance to ask whether the technology truly works for them.


Trash in, trash out: why generative AI is only as good as its data


Trash in, trash out is one of the oldest rules in data science. Generative AI doesn't get an exemption.


A large language model doesn't know facts. It learned patterns from enormous amounts of text scraped from the internet, which includes the good, the outdated and the flat-out wrong. Then it produces the most statistically likely response to your prompt. When the likely answer and the true answer are different, you get the likely one, delivered with total confidence. That's what people mean by hallucinations.


The consequences are real. In June 2023 a federal judge in New York fined two lawyers and their firm $5,000 after they filed a brief citing court opinions that ChatGPT had invented, fake quotes included.


It gets worse as the internet fills up with AI-written text. In a 2024 paper in Nature, Ilia Shumailov and colleagues trained models on the output of earlier models, generation after generation. The models lost the less common information first, then drifted into nonsense. A test that started with text about medieval architecture ended, by the ninth generation, with a list of jackrabbits. Researchers call this model collapse.


For a business, the lesson is simple. Generic inputs give you generic outputs. Wrong inputs give you wrong outputs, faster and at scale. Anything AI helps you produce is only as good as the data, expertise and judgment you put into it, which is why every QV Analytics engagement starts with the data, not the tool.


Quality over quantity: why more AI content won't save your marketing


The pitch from most AI content tools is volume. Thirty blog posts a month. A hundred social captions in an afternoon. A quarter's worth of emails before lunch.


Search engines have already priced that in. In March 2024 Google rolled out a core update and new spam policies aimed at mass-produced, unoriginal pages, and expected to cut low-quality, unhelpful content in its results by about 40% (Animalz, citing Search Engine Land). Its scaled content abuse policy goes after pages churned out in bulk mainly to game rankings, whether a person, a tool or both produced them.


In May 2026 Google updated its documentation to make clear those spam policies also cover its AI-generated answers (SEO Sherpa).


Look at it from the buyer's side, too. If your content says what every competitor's AI tool says, you've paid to sound like everyone else. The people you want to reach notice, and so do the AI answer engines that now pick which sources to cite.

What works is the material a model can't make up: your own data, your track record, a clear point of view and answers written by someone who has actually done the work.


One article with original insight beats twenty that summarize the first page of Google. That's the standard I hold this blog to, and it's the same one behind our market research and psychographics work.


The IP and copyright risks nobody puts in the sales deck


Before you hand your brand to a generator, know these three risks.


First, you may not own what it makes. The U.S. Copyright Office's January 2025 report concluded that writing prompts doesn't make you an author. Protection covers the human-made parts, like your own creative selection, arrangement or edits. In March 2025 a federal appeals court upheld that an AI can't be an author, in Thaler v. Perlmutter. If your logo, campaign or report is mostly machine-made, you may have little to stop a competitor from copying it.


Second, it may be built on someone else's work. Many models were trained on text and images collected without permission. The New York Times is suing OpenAI and Microsoft over its journalism, and in March 2025 a federal judge let the main copyright claims go forward. Until cases like this are settled, every business publishing AI output is working inside that uncertainty.


Third, you may be giving away your own IP. In 2023, Samsung engineers pasted confidential source code and a full internal meeting into ChatGPT. Samsung responded by banning staff from using generative AI tools for work, at least for a while. Depending on a tool's terms, what you type in may not stay yours.

None of this means don't use AI. It means know what you feed it, keep a human author in the loop, and read the terms before you paste in anything you'd hate to see in a competitor's hands. I'm not a lawyer, so for your specific situation, talk to one.


Is AI making us worse at thinking?


The early research says it can, depending on how you use it. "Cognitive rot" isn't a clinical term, but the pattern behind it is starting to show up in the data.


At MIT Media Lab, researchers had 54 people write essays with ChatGPT, with a search engine, or with nothing but their own heads, while EEG tracked their brain activity. The brain-only group showed the strongest, most widespread neural connectivity. The ChatGPT group showed the weakest, and it dropped over the sessions. After the first session, more than 80% of ChatGPT users struggled to accurately quote a line from the essay they'd just turned in (The Decoder). It's a small study, and it hadn't been peer-reviewed when it came out.


Microsoft Research and Carnegie Mellon surveyed 319 knowledge workers who use generative AI every week. Their finding: "confidence in AI is associated with reduced critical thinking effort" (The Register). The more people trusted the tool, the less they checked it.

And at SBS Swiss Business School, Michael Gerlich surveyed 666 people and found heavier AI use was strongly linked to weaker critical thinking, mostly through what researchers call cognitive offloading, handing the thinking over to the tool. People ages 17 to 25 leaned on AI the most and scored the lowest (PsyPost). That's a correlation, not proof that AI causes the decline.

None of these settles the question on its own. Still, three teams using three different methods landed in the same place, and when I see that in data, I pay attention. Your brain follows the same rule as a muscle: use it or lose it. Every time a tool does the reasoning and you don't check it, you skip a rep.


Never stop learning: challenging your brain is a health decision


"Never stop learning" is the motto I live by, and I mean it literally.


In the Synapse Project at UT Dallas, Dr. Denise Park's team had 221 adults ages 60 to 90 spend 15 hours a week for three months on a new activity. Only the groups learning something unfamiliar and demanding, digital photography with editing software or quilting on computer-controlled machines, improved their memory. The groups doing easy, familiar things like social outings, simple games or listening to music didn't. Park's advice: "it is important to get out and do something unfamiliar and mentally challenging" (UT Dallas).


The bigger picture points the same way. The 2024 Lancet Commission on dementia estimated that about 45% of dementia cases worldwide could be delayed or prevented by addressing 14 modifiable risk factors, and limited education is one of them (Live Science).


So here's how I use AI without letting it do my thinking:

  • Do the thinking first. Write your own outline or answer before you ask a tool for one.

  • Read every output like it came from a new intern. Check the facts, the numbers and the sources yourself.

  • Use AI to speed up work you already know how to do, not to skip learning it.

  • If you can't explain a recommendation in your own words, you're not ready to make it.

  • Pick up one hard thing outside of work, like a language, an instrument or a new sport. For me it's golf and overlanding, where every new trail is a problem to solve.


Business works the same way. I founded QV Analytics on the principle that complacency is the first step toward failure, and companies that stop learning about their market are the ones that get blindsided. Use the tools as much as you want. Just keep doing the thinking yourself.


Frequently asked questions


Is AI just a marketing buzzword?


Not entirely. Specialized AI models have produced real scientific advances, like AlphaFold's protein structure predictions and AI-supported breast cancer screening. But many products sold as AI are general text generators or basic automation with an AI label, and U.S. regulators have fined companies for overstating what their AI can do.


What does garbage in, garbage out mean for generative AI?


It means AI output is only as accurate and useful as the data and instructions behind it. Generative AI predicts likely text based on what it was trained on, so wrong, biased or generic inputs produce confident but wrong or generic results, at scale.


Can you copyright content made with AI?


In the U.S., only the human-made parts. The Copyright Office's 2025 report says prompts alone don't make you an author, and a federal appeals court ruled in 2025 that an AI can't be an author. Your own edits, selection and arrangement can still be protected.


Does using AI hurt critical thinking?


Early research suggests heavy reliance can. Studies from MIT Media Lab, Microsoft Research with Carnegie Mellon, and SBS Swiss Business School link frequent AI use with less critical thinking effort and weaker recall. Thinking first and using AI to check or speed up your own work lowers that risk.


Does Google penalize AI-generated content?


Google targets low-quality content mass-produced to manipulate rankings, whether a person or a tool wrote it. Its 2024 core update and scaled content abuse policy went after unhelpful, unoriginal pages. Original, accurate content that genuinely helps readers can still rank, however it was drafted.


Want strategy built on real data, not recycled AI output? Book a strategy call with QV Analytics.



Sources

 
 
 

Comments

bottom of page