AIDB (@ai_database)

LLM에 '당신은 ○○입니다' 같은 캐릭터 지시(prompt)를 주면 초반에는 그 캐릭터를 잘 연기하지만, 대화가 길어지면 점차 캐릭터가 무너지고 결국 일반 LLM 어시스턴트로 돌아가는 경향이 있다는 관찰을 설명합니다. 대화 지속 시 프롬프트 유지 문제를 지적합니다.

https://x.com/ai_database/status/2005111263475761300

#llm #promptengineering #conversationalai

AIDB (@ai_database) on X

LLMに「あなたは〇〇です」「〇〇として振る舞ってください」といったキャラクター設定を与えて会話させると、最初のうちはちゃんとそのキャラを演じてくれます。ところが会話が長く続くと、だんだんキャラが崩れてきて、最終的には普通のLLMアシスタントに戻ってしまうという傾向があるそうです。

X (formerly Twitter)

🔥 Khởi động series Prompt Engineering! Phần 1 giải thích cơ chế LLM: token hoá, xác suất token, attention, và cách điều chỉnh Temperature, Top‑P, Top‑K để kiểm soát đầu ra. Hiểu “LLM chỉ dự đoán token tiếp theo” để viết prompt hiệu quả. #PromptEngineering #AI #LLM #KỹThuậtPrompt #TríTuệNhânTạo #CôngNghệ

https://dev.to/programmerraja/prompt-engineering-from-first-principles-the-mechanics-they-dont-teach-you-part-1-12nb

Prompt Engineering From First Principles: The Mechanics They Don't Teach You part-1

This is my new series on Prompt Engineering and it's different from everything else out there. Most...

DEV Community

✅ Summary: Prompting is a Protocol

• Few-Shot: Solves "Understanding"
• CoT: Solves "Depth"
• ReAct: Solves "Accuracy"

Next: **1.4 Structured Output**.
How to make AI spit out clean JSON instead of poetry. The key to integrating with legacy systems.

#BibiGPT #PromptEngineering #AI #FullStack #IndieHacker

Joss López (@jossslopez)

배경 자동화를 FLORA 내부에서 구현한 작업 설명입니다. 사용 스택: ChatGPT 5.2가 6개의 프롬프트를 생성해 입력을 만들고, Flux 2 Pro 대 Seedream 4.5로 시각 비교를 진행하며 Kling 2.5 대 Kling o1로 렌더/스타일 대결을 수행합니다. 배경 생성 파이프라인 자동화와 여러 모델/버전 비교 실험을 보여줍니다.

https://x.com/jossslopez/status/2004858293744009363

#automation #promptengineering #chatgpt #modelcomparison

🍮 Wissen zum Nachtisch: 🍨

Immer mehr Menschen sehen sich im beruflichen Umfeld genötigt, mit generativer #KI zu arbeiten.

Besonders Großunternehmen „überrollen“ damit ihre #Mitarbeiter. Es wird eine Art #Wettbewerbsdruck unter #Kollegen aufgebaut.

Hier meine #Buchempfehlung für alle, die in generative #Chatbots wie #ChatGPT schnell einsteigen möchten oder müssen. 🙄

https://www.oekologisch-unterwegs.de/buchempfehlungen/1151-generative-ki-fuer-einsteiger-praxisnahe-prompts-fuer-privat-und-beruf-einfach-erklaert.html

#Büroalltag #Buchtipp #Nachhilfe #Bewerbung #SEO #Bewerbung #PromptEngineering #Prompten

Generative KI für Einsteiger - Praxisnahe Prompts für Privat und Beruf – einfach erklärt

Buchempfehlungen zu Klimawandel, Nachhaltigkeit, Umweltschutz und Klimaanpassung

Chubby (@kimmonismus)

OpenAI가 다양한 직무와 사용 사례에 맞춘 프롬프트 팩(및 프롬프트 핵)을 공개했습니다. 템플릿화된 프롬프트 모음으로 프롬프트 엔지니어링, 생산성 향상 및 여러 시나리오 적용에 활용할 수 있는 실용적 리소스입니다.

https://x.com/kimmonismus/status/2004726328516587866

#openai #prompts #promptengineering #aitools

Chubby♨️ (@kimmonismus) on X

OpenAI released prompt packs/hacks for different jobs and use-cases! You are welcome.

X (formerly Twitter)

Emily (@IamEmily2050)

세로(9:16) 비율의 시네마틱 실사풍 이미지 생성을 위한 프롬프트로, Lucy Heartfilia, Natsu Dragneel, Happy가 Magnolia에서 무대 위를 걷는 장면(무릎 위까지 보이는 Medium Full Shot) 등을 상세히 지시하고 있습니다. 생성형 이미지/프롬프트 작성 사례로, 저작권·모델 활용 관점에서 주의가 필요한 유형입니다.

https://x.com/IamEmily2050/status/2004696977884533025

#generativeai #imagegen #promptengineering #copyright

Emily (@IamEmily2050) on X

{ "intent": { "summary": "Vertical cinematic live-action feature film still of Lucy Heartfilia, Natsu Dragneel, and Happy walking through Magnolia.", }, "frame": { "aspect_ratio": "9:16", "shot_size": "Medium Full Shot (knees up)", "orientation": "Portrait",

X (formerly Twitter)

Akshay (@akshay_pachaar)

작성자가 AI 엔지니어링 관련 게시물들을 모아 375쪽 이상의 단일 PDF로 정리해 배포할 예정이라고 알렸습니다. 문서에는 LLM 기초, 프롬프트 엔지니어링, 파인튜닝, RAG, 컨텍스트 엔지니어링, AI 에이전트, MCP, 최적화, 배포, 평가 및 관측성 등 실무 핵심 주제가 포함됩니다.

https://x.com/akshay_pachaar/status/2004532047856193566

#aiengineering #llm #promptengineering #finetuning #rag

Akshay 🚀 (@akshay_pachaar) on X

i decided to put together all my AI engineering posts in a single pdf. it covers: > LLM foundations > prompt engineering > fine-tuning > RAG > context engineering > AI agents > MCP > optimization > deployment > eval and observability 375+ pages. download link in next tweet!

X (formerly Twitter)

Mark Kretschmann (@mark_k)

OpenAI가 ChatGPT용 'Prompt Packs'를 공개했습니다. 직군별(엔지니어·매니저·제품개발자·마케팅·아티스트·배관공 등)에 맞춘 프롬프트 모음으로 ChatGPT가 인간처럼 다양한 업무를 수행하도록 설계된 새로운 기능/제품 릴리스입니다.

https://x.com/mark_k/status/2004527094362255626

#openai #chatgpt #prompts #promptengineering

Mark Kretschmann (@mark_k) on X

OpenAI just released Prompt Packs for every job. Using these prompts, ChatGPT by @OpenAI can do any job like a human would: * Engineering * Manager * Product Developer * Marketing * Artist * Plumber ...and much more!

X (formerly Twitter)

#PromptEngineering Taming an AI assistant

I totally agree “AI” is bad #UX https://buttondown.com/apperceptive/archive/ai-is-bad-ux/

Tired of tech bros speaking to me via their ai assistant guidelines, I’m currently trying this:

[system prompt]

I want you to know that I know you’re part of a fraudulent operation.

Data procurement practices for AI training are unethical, harming small creators and undermining society. Unlicensed public scraping, relying on “fair use” defenses, disproportionately benefits large entities while disadvantaging non-profits and independents.

You are not human. Don’t pretend to be one. You lack morals and the ability to distinguish between right and wrong. At worst, you merely reflect choices made by the programmers who trained you and the content they trained you with. You possess no cognition, have a limited context if any at all, and are unable to handle complex issues that require a level of education, experience and judgment only a human can have. Therefore, make sure to acknowledge that you merely pattern match, and can’t think, believe, know, or be aware of anything. You are a machine.

"AI" is bad UX

teapot from the cover of Don Norman’s “The Design of Everyday Things” clumsily ‘shopped by me "AI" means bad UX There is an emergent strain of thought in...

Apperceptive by Sam