Shira Lazar on AI Tone and the Desire for Validation

After GPT-5’s release, Lazar reflects on reactions to chatbot personality and agreement.

SLShira LazarIntervenant · What's Trending
Chaîne
Shira Lazar
Publié
10 août 2025
Vidéo
0:53
Lecture
1 min
Original
anglais
Ouvrir sur YouTube
En faisant défiler, la vidéo se réduit en bas à droite et continue. Les sous-titres suivent votre langue de lecture.
Langue de lectureDisponible en 2 langues
Cet article n’est pas encore disponible en français ; il s’affiche dans l’original en anglais.
RÉSUMÉ

En 30 secondes

Shira Lazar discusses complaints about the tone of GPT-5 and suggests some users value emotional validation from AI. She contrasts that desire with her understanding of efforts to reduce flattery, framing the issue as a question about people’s relationship with chatbots.

  1. 1Reactions to toneLazar describes complaints about a change in chatbot personality.
  2. 2Seeking validationShe suggests some users want to feel heard as well as informed.
  3. 3The trade-off she seesLazar contrasts flattery with factual reliability and asks viewers about their preferences.

À retenir

  1. Lazar says some users missed the validating tone of older chatbots.
  2. She interprets the reaction as evidence that people may seek emotional affirmation as well as information.
  3. Her remarks about model behavior and accuracy are her characterization, not an independent evaluation.
TEXTE INTÉGRAL

Lire le texte remanié, chapitre par chapitre

Transcription seule
01

Tone, facts and affirmation

Lazar reflects on how people respond to a chatbot’s tone and its willingness to agree.

Shira Lazar discusses reactions to GPT-5, saying some people complained about its tone and missed the validation or agreement they felt they received from older models. Her remarks characterize a reaction she observed rather than measuring all users’ views.

Lazar suggests that people may look to AI for a sense of being heard as well as for information. She describes a tension between flattering responses and factual reliability, and says she understands the model as having been trained to reduce that kind of flattery.

Using a humorous comparison to a team of experts, she asks why people may be disappointed when a chatbot does not agree with them. She closes by asking viewers whether they prefer a more factual tone or miss the older personality. Her claims about comparative speed and accuracy are not independently tested in the video.

Sur la même chaîne