Is your "AI" just an automation script? Read this before you pay the IQ tax.
Translated from Chinese.
Introduction
AI is all but certain to become a foundational infrastructure – much like the internet. This reality brings both challenges and confusion to many industries and practitioners. Companies want to use AI to improve efficiency, but they can't tell what is and isn't actually AI. Too many businesses and professionals are being fooled by so‑called "AI" that is really just an automation tool. So this article is about: Is the "AI" you're using really AI?
What Is "AI Washing"?
"AI washing" is the practice of rebranding ordinary products and slapping an "AI-powered" label on them to ride the hype. There's a strong financial incentive behind it: studies show that startups that mention "AI" in their fundraising pitches can secure 15% to 50% more capital than those that don't.
Take Nate, for example. This company claimed its app used AI to enable "one-click shopping." In reality, the operation was mostly carried out manually by outsourced workers in the Philippines and Romania. The founder was indicted by the U.S. Department of Justice and the SEC on charges including securities fraud, facing up to 20 years in prison.
Or consider the case of South Korean home appliances. In 2025, the Korea Fair Trade Commission uncovered 20 suspected cases of AI washing. One company marketed a fan's temperature‑sensor‑based auto‑adjustment as "AI Smart Mode." Another advertised a dehumidifier's humidity‑sensing function as "AI Dehumidification Mode." In most of these cases, the products were only using basic sensor technology – no AI involved.
The Fundamental Divide: Reasoning vs. Rules
Just remember one thing: AI thinks and reasons. Automation is just logic.
Many products claim to be "AI‑powered." It's a marketing trick to make you pay for the AI label. For most people, if something is beyond their understanding, they assume it's powerful – a perfectly normal reaction. I felt the same way the first time I used VLOOKUP – I thought it was incredibly smart. But it's just logic. If you write the formula wrong, it won't find anything. That's exactly what most "AI" tools are – automation that requires strict, predetermined logic to make the whole process look smart. But there's no thinking, no reasoning, and no AI involved.
For example, some systems claim to "use AI" to capture orders, send them to a warehouse, and log them in a database. That's just automation. Some AI summarization tools may actually use AI – but they can also be built with scripts: capture voice, convert it to text with speech‑to‑text, and summarize the output. Also automation. And then there are robot vacuums that package infrared/laser collision detection as "AI path planning," or pixel‑by‑pixel image comparison as "AI vision model recognition." All of these are just logic that any script can handle. Add the word "AI" and the price goes up by a few hundred bucks. Still automation.
Marketing decade‑old technology as "AI" – simply because the public isn't yet sensitive enough to tell the difference – is the biggest scam of this early AI era.
AI Thinks, Automation Only Iterates
If you want to tell them apart, remember: AI thinks and reasons. What AI gives you is a probability.
Take the robot vacuum example. If a vacuum were truly AI‑powered, with the right toolset, it could clean every inch of your home thoroughly:
Detects unknown object → it's cat poop → pick it up with a small shovel → drop it in the bathroom
Detects a corner → use tweezers + small scraper
Detects liquid → it's oil → no tool available, skip it
Detects liquid → it's water → mop normally
To program a robot to handle every single type of stain, you'd have to cram it with countless "if‑then" rules – eventually, the machine would choke. You'd even see it freeze in front of a single strand of hair, running through options one by one, trying to figure out what it's dealing with. Which is why, in practice, it just drags everything – poop and oil alike – and smears it evenly across every inch of your floor. Go ahead and search for those videos.
The Cost of Probability: Why You Can't Use AI in High‑Stakes Environments
Take autonomous driving. Until the next generation of models arrives – the truly next generation – there won't be AI‑driven cars. Because AI thinks, but it's not smart enough, and it suffers from hallucination and sycophancy. It's unreliable.
A 1% hallucination in text is just noise. A 1% probability of losing control in a safety‑critical system is 100% deadly.
Here's an example: an automation script processes a braking command – input A always produces a 100% deterministic mechanical brake. AI, on the other hand, works on probability sampling. If a pedestrian merely glances toward the road, the AI might notice it, start processing, and because there's no hard‑coded rule to fall back on, it might over‑amplify the risk of a "pedestrian crossing and collision," then slam the brakes for no reason at highway speeds. That's disastrous for the driver.
Choosing by the Boundaries of Rules: Use AI to Build Automation
AI is not a universal cure‑all – even though it's genuinely useful. But in clearly rule‑bound environments – like assembly lines or traffic – logic is far more stable than AI.
Automation: If → not → next → not → next → not → next → execute
AI: If → what's this, let me think → execute
Automation iterates through options. AI overthinks. On the road – a place with vague requirements, too many rules, and too many exceptions – AI has almost no chance. Similarly, on an assembly line – clear requirements, clear rules, and few exceptions – there's no need to waste AI compute on it.
AI's real strength is in environments where requirements are clear but rules are fuzzy. Let AI use any possible approach to get your work done. Or better yet – use AI to write scripts and programs based on your ideas. You get the convenience of AI and the accuracy and stability of deterministic execution.
That's what I do. It's more accurate, faster, and cheaper (at least I'm not burning tokens). AI's role here is to fill the gap in my ability to build automation tools.
So don't stress too much about whether you're using AI. Maybe AI simply doesn't fit your work environment. Automation might be your best option.
Summary
AI and automation are not substitutes for each other. There's an old Chinese saying: "Whether it's a black cat or a white cat, it's a good cat if it catches mice."
But as users, we should be careful – don't get charged an "IQ tax" just because someone says their product is "AI‑powered."
Top comments (0)