Academic English revision

AI Inflated Your Results? How to Kill Ambiguous Modifiers

AI expansion creates ambiguous modifiers that warp scientific findings. Precision beats flow every time in the Results section.

It’s 2 AM. You’ve just used AI to expand a dry sentence about your experimental data into a sophisticated paragraph. It feels professional, so you hit submit. The next morning, your professor marks it up: “Ambiguous modifier—does ‘significant’ refer to the effect or the sample?” Your stomach drops. This is the classic failure mode of AI-assisted writing: modifiers going rogue in scientific contexts.

AI loves to pad sentences with complex clauses to sound 'academic.' In a Results section, this is dangerous. When you describe an experiment, a misplaced adjective or adverb can completely flip the meaning of your data. You are reporting facts, not writing poetry. ❌ The observed reduction in error rates, which was significant after the update, led to faster processing. ✅ After the software update, error rates dropped significantly, leading to faster data processing.

Many students think the Results section is just a list of numbers. Wrong. It’s a narrative, but it must be a linear one with zero logical gaps. When you expand sentences, the modifier must sit right next to what it describes. If your variable is 'time,' your adverb must cling to the verb associated with time, not drift over to 'efficiency.' This is physics, not style. ❌ The rapid increase in temperature and humidity caused the machine to stop. ✅ The machine stopped due to a rapid increase in both temperature and humidity.

The 'A and B' trap is the most common ESL error I see. AI loves conjunctions to add length. But if you modify only 'A' and forget 'B,' or place the modifier at the start without scope, readers will guess. In empirical research, guessing is a fatal flaw. ❌ We observed the effect of the drug on both adults and children, resulting in fewer side effects. ✅ The drug resulted in fewer side effects when administered to both adults and children.

How do you fix this? Stop obsessing over grammar labels. Use the 'noun-scratch' test: cross out all nouns and look at what remains. If the logic breaks, your modifiers are detached. I don’t do this manually anymore; I run the draft through easydue to shatter those tangled clauses and re-weld them onto the correct logical nodes. Save your mental energy for checking your p-values, not agonizing over word order. ❌ The increased sample size and the new algorithm improved accuracy. ✅ Accuracy improved because of both the larger sample size and the adoption of a new algorithm.

There’s also the invisible killer: passive voice with vague subjects. AI is addicted to "It was found that..." In your Results, this 'it' is a black hole for ambiguity. What does 'it' refer to? The data? The hypothesis? The trial? ❌ It was reported that the model failed to converge when training data was limited. ✅ The model failed to converge during training on datasets with fewer than 500 samples.

Remember: in academic writing, clarity is king. 'Flow' means nothing if it creates confusion. Your professor doesn’t care about your vocabulary; they care that you’re intelligible on the first read. If AI expands your sentences, you are responsible for tightening their logic. Don’t let modifiers ghost around the page. Pin them to the right noun. That’s what science is: cold, precise, and unforgiving of ambiguity.