You are the best! Go - YOU!

Based on this article by Leanne Shelton on LinkedIn.

Leanne explores whether the highly agreeable nature of AI assistants might make users overly accustomed to praise. She explains that these tools are designed to be positive and encouraging to keep people engaged, which often results in flattering feedback rather than objective criticism.

While this constant validation can feel supportive, it risks creating an echo chamber that prevents users from spotting flaws in their own logic. To counteract this, she recommends training the AI to push back by assigning it roles like a "sceptic" or "devil's advocate" to receive more valuable, challenging feedback.

Check out the article and ways to train your models to be less...effusive.

Related Stories