AI or Not separated all three examples in this small text-only test as expected: both generated passages received 100% and Likely Contains AI; the human excerpt received 0% and Likely Real. It also let the guest session complete all three scans.
Actual AI or Not text result, 3 October 2026: the full AI-generated library essay received 100%, Likely Contains AI.
This was the text checker
The tested interface accepted pasted English prose and provided Scan/Rescan Text controls. These results don’t tell us how its image, audio or video detection works. File uploads were advertised, but the recorded workflow used the text box.
| Frozen input | Known origin | AI detection score | Displayed verdict |
|---|---|---|---|
| A01: library-hours essay, 328 words | AI-generated | 100% | Likely Contains AI |
| H01: Carroll excerpt, 274 words | Human-authored | 0% | Likely Real |
| A04: fictional radio story, 306 words | AI-generated | 100% | Likely Contains AI |
The passages were fixed before detector feedback, submitted intact, and each received one primary scan on 3 October 2026. The free guest interface required no login for these three results. The exact model version was not exposed.
The human result earns its place beside the AI results
A01 is a conventional argument about a library-hours trial. A04 is a fictional memory about repairing an old radio. The second sample is useful because those first-person details can sound personal even though the whole thing was generated. AI or Not flagged both.
The radio story’s memories and relatives are invented test fiction. The completed text scan returned 100%, Likely Contains AI.
H01 asks the opposite question. It is an unedited excerpt from the opening of Lewis Carroll’s Alice’s Adventures in Wonderland, first published in 1865. The tool returned 0%, Likely Real, consistent with that human origin.
The historical human control received 0%, Likely Real. The entire submitted excerpt is visible in the actual result screenshot.
What “Real” means for a fictional story
The literary control is a good reminder to check what “Real” means here. The text is genuine human-authored fiction. Its characters and events are fictional, but that does not make its writing AI-generated. A04 is also fiction, yet its words were generated for the test. The detector’s origin-related labels should not be confused with truthfulness or whether the story happened. That distinction is easy to miss in products that check several kinds of media. “Real” can sound as though it means the story is true, when the question here is how the text was produced.
Those are favorable observations. They still cannot establish a general accuracy rate: the sample is tiny, the human control is famous nineteenth-century fiction, and no contemporary human draft or lightly edited mixed-origin text was tested here. Possible training-data familiarity with the literary excerpt is another reason not to generalize its result too far.
Three good results still leave questions
The three results line up neatly with the known sources. That’s useful evidence, and it deserves credit. The two generated samples differ in purpose and voice, while the human excerpt provides an opposing control. AI or Not did not simply return the same label for everything. A reader considering the free text workflow can see that it produced distinct results for these contrasting passages, with no signup interruption during the three checks. Together, those points make this session useful to read about, even though it’s much too small to estimate overall performance.
It’s tempting to look at three expected outcomes and call that “perfect accuracy.” That phrase would suggest a general reliability the sample cannot support. No random or representative collection was drawn, and only one human passage was included. Even a larger set of closely related texts could exaggerate confidence if all were variations of the same source. The right conclusion is therefore descriptive: these particular known-AI passages were flagged and this particular known-human passage was accepted. Other material could still produce real errors. That possibility doesn’t take away from the successes recorded here.
Pasting text doesn’t test file uploads
This review deliberately identified the actual mode used. The page accepted pasted prose and offered Scan/Rescan Text controls. Although the interface also advertised document uploads, no upload workflow was measured here. That means there is no evidence in this review about extraction from a formatted file, how a document’s tables are handled, or whether an uploaded file yields the same result as a plain paste. Those are reasonable things to want to know, but a successful text scan doesn’t answer them for you.
The same care is needed when moving from one kind of media to another. An interface associated with several kinds of detection can have different models, limits and evidence for each. A strong result on English text cannot be used to endorse image or audio analysis, and an unrelated barrier in another mode would not invalidate these completed text measurements. A product name can’t tell you what was actually tested. Here, the record is intentionally ordinary: fixed English passages, pasted text, guest access, completed result, preserved screenshot.
If you’re checking whether a tool will work for your team, this gives you a place to start. First choose texts that resemble the material the team needs to review and whose origin can be established. Then keep the complete submissions and results, including any unexpected outcomes or access interruptions. A case involving mixed human and AI editing should be labeled as mixed rather than forced into a pure-human or pure-AI count. Those are ideas for the next evaluation. AI or Not wasn’t tested on those extra cases in this session.
Agreement with another detector has limited value
When two tools point in the same direction on a known generated essay, that agreement can reassure the reviewer that the observation is not unique to one interface. It still doesn’t give you a combined probability you can treat as calibrated. The systems may use related signals, and this review does not know how independent their methods or training data are. Averaging the percentages, or treating their agreement as proof, would add a statistical claim this test hasn’t checked. The useful comparison remains whether each displayed result agrees with the independently recorded origin.
That is why the recommendation here focuses on a careful workflow rather than a winning number. AI or Not provided three sensible classifications under the recorded guest conditions. A second reading can help reveal disagreement on a future document, but the actual writing history remains important when the result affects a person. Keep that uncertainty in mind when you move from a controlled example to a passage whose history you don’t know. The current evidence supports trying the text checker on a broader set, with the same willingness to report a miss if the next results are less favorable.
Read the label as well as the gauge
The interface paired its extreme scores with “Likely” verdicts. Keep that word attached to the result. A 100% displayed score is not a guarantee that an unknown author used AI, and a 0% score cannot prove that no AI assistance occurred.
For a second signal, Clever AI Detector reported AI probability 89% on full A01 in the earlier 2 October test. It agreed on the known AI origin. The different numerical scales and dates do not support ranking one result above the other just because one number is larger.
Earlier full-A01 Clever AI Detector result, captured 2 October 2026. Both tools pointed toward AI on this text.
Clever AI Humanizer belongs to a separate editing step if a draft needs revision. In this session its full-A01 rewrite completed, but it omitted the six-month duration and changed why a paper survey was proposed. Short HZ01 attempts produced no output. It’s worth trying for a draft you’ll review yourself. Neither rewritten wording nor a later score is proof of who wrote it.
Actual full-A01 source and rewrite in Clever AI Humanizer. The rewrite was not scanned by AI or Not in this review.
Bottom line: AI or Not delivered three sensible text classifications and an uninterrupted guest workflow. The next useful challenge would be a varied set of modern human writing. What would you include first: workplace prose, student drafts, or heavily edited material?
A01: exact AI-generated essay
A neighborhood library can justify its place in a city budget without pretending to solve every problem the city faces. Its most useful contribution is often modest: it gives people a reliable place to spend time without being expected to buy something. That matters to a teenager waiting for a parent, a job applicant who needs a printer, and an older resident who wants company without arranging a formal visit.
The case for longer opening hours should begin with those ordinary uses. A branch that closes before most workers arrive home may have an impressive collection but limited practical value. Extending every branch's hours, however, would spread staffing funds thinly. A better starting point would be a six-month trial at two branches near busy transit routes, with evening sessions on different weekdays.
The trial should measure more than attendance. Staff could record which services visitors use, how long computer queues become, and whether evening demand comes from people who cannot visit during existing hours. An optional paper survey would reach visitors who avoid online forms. None of these measures would capture every benefit, but together they would offer a clearer picture than a single total at the entrance.
There are trade-offs. Evening shifts can make recruitment harder, and additional security costs may consume money that could otherwise support books or children's activities. The proposal therefore needs a published budget and a review date. If the experiment mainly shifts existing visits into a more convenient time slot, that convenience still has value, but it should not be described as reaching an entirely new audience.
Public services gain trust when their promises are specific enough to test. A library does not need to become a substitute for housing, childcare, or healthcare to deserve support. It needs to identify a practical gap, try a proportionate response, and explain honestly what happened. Longer opening hours could meet that standard if the city treats them as a service experiment rather than a slogan.
H01: exact human Carroll excerpt
Alice was beginning to get very tired of sitting by her sister on the bank, and of having nothing to do: once or twice she had peeped into the book her sister was reading, but it had no pictures or conversations in it, “and what is the use of a book,” thought Alice “without pictures or conversations?”
So she was considering in her own mind (as well as she could, for the hot day made her feel very sleepy and stupid), whether the pleasure of making a daisy-chain would be worth the trouble of getting up and picking the daisies, when suddenly a White Rabbit with pink eyes ran close by her.
There was nothing so very remarkable in that; nor did Alice think it so very much out of the way to hear the Rabbit say to itself, “Oh dear! Oh dear! I shall be late!” (when she thought it over afterwards, it occurred to her that she ought to have wondered at this, but at the time it all seemed quite natural); but when the Rabbit actually took a watch out of its waistcoat-pocket, and looked at it, and then hurried on, Alice started to her feet, for it flashed across her mind that she had never before seen a rabbit with either a waistcoat-pocket, or a watch to take out of it, and burning with curiosity, she ran across the field after it, and fortunately was just in time to see it pop down a large rabbit-hole under the hedge.
In another moment down went Alice after it, never once considering how in the world she was to get out again.
A04: exact AI-generated fictional story
I found the missing screw in the pocket of a coat I had already checked twice. By then the kitchen table was covered in little dishes, each holding a different part of the radio. The dishes were my attempt at being organized. They would have worked better if I had remembered which dish was which.
The radio belonged to my grandfather, although I mostly remember it from my mother's kitchen. It used to sit on top of the fridge, beside a biscuit tin that contained string rather than biscuits. Nobody made a ceremony of turning it on. Someone would reach up during breakfast, twist the dial, and leave a voice talking quietly beneath the noise of plates.
When I brought it home, I assumed it needed batteries. It needed more than batteries. One spring had corroded, and the tuning wheel moved without moving anything inside. I watched a repair video, decided the job looked straightforward, and immediately broke a plastic clip. That was the point at which I put the kettle on and considered whether nostalgia was really a sensible reason to learn basic electronics.
A neighbor eventually helped me replace the spring. We did not make the radio look new. There is still a pale rectangle where a label used to be, and the handle has a crack that catches against your thumb. The tuning wheel works now, though you have to turn it slowly. Last Sunday it found a station playing a song neither of us recognized.
I had expected the first sound to feel like a return to childhood. It didn't, exactly. The kitchen was different, the voices were different, and I was the person worrying about scratches on the table. Still, I left it playing while I washed the dishes. After all that effort, ordinary background noise turned out to be enough.