Lesson 142 · 2 min
AI-generated reviews
Product and restaurant reviews can now be written by AI in bulk, so read reviews for specific details and patterns instead of star counts.
Sellers can generate hundreds of natural-sounding five-star reviews, or attack a rival with one-star ones. Paid review groups make it worse.
Fake reviews tend to be vague and enthusiastic ("Great product, highly recommend!"), appear in a burst on the same few days, and repeat similar phrases.
Real reviews usually mention specific details: how long the delivery took, a small flaw, how the size fits. Photos taken by buyers and mixed three- and four-star reviews are often the most useful.
Check the reviewer's profile, sort by newest, and read the critical reviews. Buy from shops with a return policy you can use.
Do it now
Read reviews like a detective
- 1. Open a product you are thinking of buying.
- 2. Read five three-star reviews and five one-star reviews.
- 3. Look at the dates of the five-star reviews for suspicious bursts.
Here are some reviews of a product I want to buy: [PASTE REVIEWS]. Group the common points, note any that sound generic or repeated, and tell me what real buyers seem to agree on. Be clear that you cannot prove which reviews are fake.
Quick check
Words from this lesson
- Generative AI
- AI that produces new text, images, audio, video or code, rather than only sorting or scoring existing things.
- LLM
- An AI trained on huge amounts of text to predict the next word, which lets it write, summarise, translate and chat.