Conversation

Notices

Embed this notice
Mathieu Triclot (mtriclot@aleph.land)'s status on Monday, 04-Dec-2023 23:10:17 JST Mathieu Triclot

Eliza bat GPT3.5 au test de Turing : https://arstechnica.com/information-technology/2023/12/real-humans-appeared-human-63-of-the-time-in-recent-turing-test-ai-study/ | le style formel des réponses pose problème, même en adaptant les prompts.
In conversation Monday, 04-Dec-2023 23:10:17 JST from aleph.land permalink
Attachments
1. Domain not in remote thumbnail source whitelist: cdn.arstechnica.net
  
  1960s chatbot ELIZA beat OpenAI’s GPT-3.5 in a recent Turing test study
  
  from @benjedwards
  
  AI chatbot deception paper suggests that some bots (and people) aren't very persuasive.
- Embed this notice
  Mathieu Triclot (mtriclot@aleph.land)'s status on Monday, 04-Dec-2023 23:10:14 JST Mathieu Triclot
  in reply to
  
  Variante : chatGPT vs stack overflow. Les utilisateurs préfèrent les réponses de chatGPT, car assertatives, alors que 52% d'erreurs.
  https://arxiv.org/abs/2308.02312
  In conversation Monday, 04-Dec-2023 23:10:14 JST permalink
  Attachments
  1. Domain not in remote thumbnail source whitelist: static.arxiv.org
    
    Who Answers It Better? An In-Depth Analysis of ChatGPT and Stack Overflow Answers to Software Engineering Questions
    
    Over the last decade, Q&A platforms have played a crucial role in how programmers seek help online. The emergence of ChatGPT, however, is causing a shift in this pattern. Despite ChatGPT's popularity, there hasn't been a thorough investigation into the quality and usability of its responses to software engineering queries. To address this gap, we undertook a comprehensive analysis of ChatGPT's replies to 517 questions from Stack Overflow (SO). We assessed the correctness, consistency, comprehensiveness, and conciseness of these responses. Additionally, we conducted an extensive linguistic analysis and a user study to gain insights into the linguistic and human aspects of ChatGPT's answers. Our examination revealed that 52% of ChatGPT's answers contain inaccuracies and 77% are verbose. Nevertheless, users still prefer ChatGPT's responses 39.34% of the time due to their comprehensiveness and articulate language style. These findings underscore the need for meticulous error correction in ChatGPT while also raising awareness among users about the potential risks associated with seemingly accurate answers.
- Embed this notice
  Mathieu Triclot (mtriclot@aleph.land)'s status on Monday, 04-Dec-2023 23:10:16 JST Mathieu Triclot
  in reply to
  
  En revanche sur les visages, ça passe : les visages engendrés par IA sont jugés plus humains que les originaux
  https://arstechnica.com/information-technology/2023/11/people-think-white-ai-generated-faces-are-more-real-than-actual-photos-study-says/
  In conversation Monday, 04-Dec-2023 23:10:16 JST permalink
  Attachments
  1. Domain not in remote thumbnail source whitelist: cdn.arstechnica.net
    
    People think white AI-generated faces are more real than actual photos, study says
    
    from @benjedwards
    
    'Hyperrealism' bias may impact medicine and law enforcement. Can you spot the fakes?
  Haelwenn /элвэн/ :triskell: likes this.
  
  Haelwenn /элвэн/ :triskell: repeated this.

Public

Conversation

Notices

Feeds