home.social

#promptdesign — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #promptdesign, aggregated by home.social.

fetched live
  1. Treating AI prompts not as one-off queries, but as an iterative, human-centered design process transforms garbage-in, garbage-out, it’s empathy-in, quality-out.
    #DesignThinking #PromptEngineering #GenerativeAI #UXDesign #AI #PromptDesign #IterativeDesign

  2. Developers are becoming directors, not typists. As large language models turn into coding agents, the real skill shifts to prompt design and orchestration. Simon Last explains how AI productivity tools like Codex reshape software development and why open‑source collaboration matters more than ever. #AICodingAgents #PromptDesign #SoftwareDev #MachineLearning

    🔗 aidailypost.com/news/developer

  3. Eine Einführung in das Prompt-Design mit den wichtigsten Konzepten und Ansätzen - von klaren Formulierungen bis zu fortgeschrittenen Techniken. Hier ein Überblick:
    arxiv.org/abs/2401.14423

    #KI #Prompt #Promptengineering #Promptdesign

  4. Too many people believe that simply providing a prompt to a language model will always result in accurate and reliable responses.

    🚫 The Reality:

    While prompts are a crucial part of guiding language models, it's a common misunderstanding to think that any prompt will consistently produce accurate and meaningful outputs. Language models are complex and can be sensitive to the way prompts are formulated. Evaluating and refining prompts is essential for achieving reliable results.

    Here's why:

    1. 𝗩𝗮𝗿𝗶𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗶𝗻 𝗠𝗼𝗱𝗲𝗹 𝗥𝗲𝘀𝗽𝗼𝗻𝘀𝗲𝘀: Language models can generate a wide range of responses based on slight variations in prompts. A poorly formulated prompt can lead to irrelevant or nonsensical answers.

    2. 𝗕𝗶𝗮𝘀 𝗮𝗻𝗱 𝗠𝗶𝘀𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻: Language models amplify biases present in the training data. Without careful prompt design and evaluation, they may produce biased or inaccurate information.

    #AI #LanguageModels #PromptDesign #Career #Tech

  5. 🧠 Martedì parteciperò a un incontro promosso dalla Camera di Commercio di Roma dedicato all'#AI e agli algoritmi generativi. 
    🦾 Il corso è destinato alle PMI, e sarà l’occasione per avvicinarsi a sistemi basati su AI e Machine Learning, e per capire il funzionamento dei modelli di linguaggio (es. #GPT4, #ChatGPT), e le basi di prompt design.
    📆 Martedì 25 dalle 10 alle 13, online.
    📄 Verrà rilasciato l'attestato di partecipazione.

    #LLM #PromptDesign

    👇 per iscriversi 👇

    formacamera.it/attivita/ciaag0

  6. 🧐 A cosa servono le Catene di Pensiero (#CoT) nel mondo dei #LLM? Perché rappresentano una tecnica vincente? 
    💡 La risposte è semplice. Servono a creare un contesto solido per il modello, e questo rappresenta la chiave per ottenere il massimo rendimento d questi algoritmi.

    #AI #IntelligenzaArtificiale #GPT4 #PromptEngineering #PromptDesign 

  7. 🖼️ Queste immagini sono state generate usando #Midjourney V5.2 partendo da #prompt della versione precedente processati attraverso il comando /shorten.
    🧐 Questa novità del modello è sottovalutata, ma secondo me offre un grande assist per comprendere come vengono interpretate le richieste.
    🔗 Il post che riporta anche i prompt: bootcamp.uxdesign.cc/midjourne

    #AI #TextToImage #LLM #PromptDesign #PromptEngineering

  8. #introduction
    Hi, I'm a PhD student studying #AIethics at Uni. Sydney 🇦🇺 in the Philosophy Science dept.

    Since late 2021 I've been a Student Researcher at Google Research. I started in #EthicalAI under the #ResponsibleAI umbrella and now work with people on a few different teams.

    I research values in #LanguageModels using a lens of #ValuePluralism.
    My grounding is social science, philosophy, & cybernetics.

    #PromptDesign #Benchmarks. #ResponsiblePromptDesign

    I ❤: 🐶 ⛷ 🏝 🌏