AI answers to health questions read at a grade 11.3 level, close to a high school senior's, the hardest of any topic and four and a half grades above recipe answers. On Gemini, health answers read at grade 12.6. The average AI answer reads at grade 9.3.
Analysis of 25,427 AI answers to 2,984 questions · answers from ChatGPT, Gemini, Perplexity and Google's AI features, September 2026
Key Findings
- 1.AI answers to health questions read at a grade 11.3 level, the hardest of any topic, against grade 6.8 for recipe answers.
- 2.Gemini's answers to health questions read at a grade 12.6 level, the highest of any AI engine on health questions.
- 3.The average AI answer reads at a grade 9.3 level, about what a US ninth grader can read comfortably.
- 4.Gemini writes the hardest AI answers at grade 10.4 and ChatGPT the easiest at grade 8.3, a gap of two school grades.
- 5.Google AI Overviews read at a grade 8.5 level, one and a half grades easier than Google AI Mode at grade 10.
Summary
A parent looks up whether a child's rash needs a doctor. The AI answer is accurate, but it is written in long sentences full of clinical terms. Ask the same engine how to make pancakes and the answer reads like a note from a friend.
That gap matters because the topics where AI writes hardest are the ones where readers most need to understand the answer and act on it: health, mental health and legal questions. It also matters to anyone writing content AI will use. The reading level of an AI answer shows how the engine pitches a topic, and pages written far above it are harder to quote as they are.
AI answers to health questions read at a grade 11.3 level, the hardest of 31 topics, and Gemini's health answers at grade 12.6. The average AI answer reads at grade 9.3, and recipe answers at grade 6.8, a spread of more than four school grades between the easiest and hardest topics.
What we measured
On September 23, 2026 we asked 2,984 everyday questions in 31 topics, from health and legal to recipes and gardening, to five AI engines: ChatGPT, Gemini, Perplexity, Google AI Mode and Google AI Overviews (the AI summary above Google's search results). We scored the 25,427 answers of 30 words or more.
Each answer was scored with the Flesch-Kincaid grade level, a standard readability formula that turns sentence length and word length into a US school grade. Grade 6 text suits most 11-year-olds; grade 12 is the level of a high school senior. A higher grade means longer sentences and longer words. We averaged the grades by engine and by topic.
The average AI answer reads at a ninth-grade level
AI answers are pitched at about a ninth-grade reading level, the level of a high school freshman. That is readable for most adults, though harder going for anyone reading quickly on a phone or in a second language.
The average hides a wide spread. The topic moves the grade by more than four levels, from 6.8 for recipes to 11.3 for health, and the engine moves it by two, from 8.3 on ChatGPT to 10.4 on Gemini. Put together, a reader can get a sixth-grade answer or a twelfth-grade answer depending on what they ask and where.
Health answers are the hardest to read
Health (11.3), mental health (11), business (10.8) and legal (10.7) answers are the most complex. These topics carry medical, legal and financial terms that push word length up, and AI engines explain them in long, qualified sentences. Gemini's health answers read at grade 12.6, and on business, legal and mental health questions Gemini also reaches grade 12 or above.
Source: Suff Digital analysis of 25,427 AI answers to 2,984 questions · answers from ChatGPT, Gemini, Perplexity and Google's AI features, September 2026
| Topic | ChatGPT | Gemini | AI Mode | AI Overviews | Perplexity | All engines |
|---|---|---|---|---|---|---|
| Health | 10.1 | 12.6 | 12.3 | 10.3 | 10.5 | 11.3 |
| Mental health | 10.4 | 12.1 | 11.7 | 9.9 | 10.9 | 11 |
| Business | 9.5 | 12.3 | 11.9 | 9.8 | 10.4 | 10.8 |
| Legal | 9.2 | 12 | 11.8 | 9.7 | 9.8 | 10.7 |
| Parenting | 9.2 | 11.7 | 11.7 | 9.1 | 9.8 | 10.4 |
| Cars | 9.6 | 11.5 | 10.8 | 9.3 | 10.5 | 10.4 |
| Insurance | 9.6 | 10.9 | 10.1 | 9.3 | 10.4 | 10.1 |
| Money | 8.2 | 10.3 | 10.3 | 8.1 | 9.5 | 9.3 |
| Travel | 8.2 | 10.3 | 9.7 | 9 | 9.6 | 9.4 |
| Tech | 6.9 | 9.8 | 9.6 | 8.1 | 8.8 | 8.6 |
| Taxes | 7.4 | 9.4 | 9.3 | 8.3 | 8.6 | 8.6 |
| Software | 6.7 | 10.2 | 9 | 7.4 | 8.9 | 8.4 |
| Garden | 7.4 | 8.5 | 8.4 | 6.6 | 9.2 | 7.9 |
| Beauty | 7.2 | 8.5 | 8.2 | 6.9 | 8.5 | 7.8 |
| Recipes | 6 | 7.2 | 7.3 | 6.3 | 7.6 | 6.8 |
Recipes (6.8), beauty (7.8), gardening (7.9) and software (8.4) read most easily. ChatGPT's recipe answers read at grade 6, the easiest in the table. The pattern puts the hardest reading exactly where the stakes are highest: a reader worried about a symptom or a legal problem gets a denser answer than one choosing a recipe.
Gemini writes hardest, ChatGPT simplest
Gemini (10.4) and Google AI Mode (10) both write at grade 10 or above. Perplexity sits near the average at 9.4, and Google AI Overviews (8.5) and ChatGPT (8.3) are the most readable.
Source: Suff Digital analysis of 25,427 AI answers to 2,984 questions · answers from ChatGPT, Gemini, Perplexity and Google's AI features, September 2026
By topic, ChatGPT and Google AI Overviews trade places as the easiest engine. ChatGPT drops to grade 6 on recipes and 6.7 on software, while Gemini reaches grade 12 or above on health, business, legal and mental health.
Google's two AI features read very differently
Google AI Overviews read at grade 8.5 and Google AI Mode at grade 10, even though both come from Google. The gap shows up in nearly every topic: on health, AI Overviews score 10.3 against 12.3 for AI Mode; on parenting, 9.1 against 11.7; on money, 8.1 against 10.3.
AI Overviews are built mostly from short bullet points, and short list items lower the grade. AI Mode writes the longest answers of any engine, at 315 words on average, with room for longer, more qualified sentences. A reader who moves from an AI Overview to AI Mode for more detail gets it at a noticeably higher reading level.
What this means for content writers
Write at or below the level AI writes. Short sentences and common words make your text easier to quote as it is, and they suit readers who arrive from an AI answer pitched at grade 8 or 9. If your page on a topic reads several grades above ChatGPT's answer to the same question, trim it.
Simplify hardest where AI is hardest. In health, legal and financial content, define terms in plain words the first time you use them and keep one idea to a sentence. A page that explains a medical or legal point in grade 8 language offers an AI engine a clearer source than the grade 12 answer it would otherwise write.
Then test a few key pages with a readability checker and compare them with the AI answers for the same questions. Our seo copywriting services rewrite pages to the reading level AI engines and readers both handle easily.
Embed this research
Paste this on your site to embed the charts. It links back to the source automatically.
<iframe id="sd-ai-answers-reading-grade-level" src="https://www.suffdigital.com/embed/data-studies/ai-answers-reading-grade-level" width="100%" height="600" style="width:100%;border:1px solid #E5E7EB;border-radius:12px" loading="lazy" title="AI Answers to Health Questions Read at a Grade 11.3 Level vs 6.8 for Recipes - Suff Digital"></iframe>
<script>window.addEventListener("message",function(e){if(e&&e.data&&e.data.sdEmbed==="ai-answers-reading-grade-level"&&e.data.height){var f=document.getElementById("sd-ai-answers-reading-grade-level");if(f){f.style.height=e.data.height+"px";}}});</script>
<p style="font:14px/1.5 system-ui,sans-serif">Source: <a href="https://www.suffdigital.com/resources/data-studies/ai-answers-reading-grade-level">AI Answers to Health Questions Read at a Grade 11.3 Level vs 6.8 for Recipes - Suff Digital</a></p>
Cite this study
Suff Digital. (2026). AI Answers to Health Questions Read at a Grade 11.3 Level vs 6.8 for Recipes. https://www.suffdigital.com/resources/data-studies/ai-answers-reading-grade-level
AI Answers to Health Questions Read at a Grade 11.3 Level vs 6.8 for Recipes - Suff Digital - https://www.suffdigital.com/resources/data-studies/ai-answers-reading-grade-level
Frequently asked questions
Related studies
- AI SEOGoogle AI Mode Writes 315 Words per Answer vs 141 for Perplexity
- AI SEOGemini Answers 77.1% of Health Questions Without Citing a Source vs 2.4% for ChatGPT
- AI SEOAI Health Answers Cite YouTube More Often Than the NIH: 28.4% vs 26.7%
- AI SEO8.1% of AI Health Answers Use a Standard 'See a Doctor' Phrase