Ask five learning designers whether an AI tutor should say "that's incorrect" versus something softer, and you'll get five different answers with real conviction behind each one. It's a smaller question than it sounds — and it turns out to matter more than most interface decisions these platforms make.

The case for softening it

Designers who favor gentler framing point to research on math anxiety and stereotype threat: students who already doubt their ability in a subject can disengage entirely after a blunt correction, even one meant constructively. Softer framing — "let's look at this part again" rather than "that's wrong" — keeps a hesitant student in the exchange rather than triggering a shutdown response.

The case for bluntness

The opposing camp argues that excessive softening can obscure whether an answer was actually wrong at all, particularly for older students and professional learners who report finding euphemistic feedback patronizing or, worse, genuinely unclear about what to fix.

If a nursing student gives the wrong dosage calculation, I don't want the tutor to say 'let's revisit this.' I want it to say 'that's incorrect, and here's why it matters.'— Learning designer, professional/vocational education

Where the field seems to be landing

The emerging consensus among the designers we spoke with isn't a single tone — it's tone matched to stakes and audience. Early-stage and younger learners, and any subject with documented anxiety effects, lean toward softer framing. High-stakes technical or professional training leans toward direct, unambiguous correction, because the cost of an unclear correction is higher than the cost of a blunt one.

What nearly everyone agreed on: the tone should be a deliberate design choice a platform can explain, not an accidental byproduct of how the underlying model happens to phrase things by default.