My #AIText #Moodle question type uses an external AI/LLm to evaluate student responses. I have developed a branch which allows the creation of multiple test responses for use when editing a question. You can see the branch and test it out at
https://github.com/marcusgreen/moodle-qtype_aitext/tree/response_test
It is beta code but I am fairly confident it works as expects and I intend merging it into the main branch in the near future. Feedback would be apprciated
→ ‘An Overwhelmingly Negative And Demoralizing Force’: What It’s Like Working For A Company That’s Forcing AI On Its Developers
https://aftermath.site/ai-video-game-development-art-vibe-coding-midjourney
“Bradley says his boss will just keep #prompting an AI for #images until he finds one he likes, and then the art team will have to #backwards #engineer the whole thing to make it work.
"He doesn't know that the important thing isn't just the end #result, it's the #journey and the questions you answer along the #way"”
L’#IA exacerbe la relation ambivalente entre clients et automates – FredCavazza.net
https://fredcavazza.net/2025/04/08/lia-exacerbe-la-relation-ambivalente-entre-clients-et-automates
La conclusion de cet article est donc logiquement à contre-pied de ce que vous pouvez lire ou entendre à droite et à gauche : l’urgence pour les entreprises n’est pas d’adopter l’IA au plus vite (former les salariés au #prompting) ou d’agentiser leurs processus mais de réduire la dette technique et de lutter contre la dispersion des informations et données
Ollama v0.5.0 pre-release introduces structured JSON output, allowing developers to constrain model responses to a specific JSON schema.
So you no longer need time-consuming prompt engineering for output reliability and consistency, which is pretty awesome.
You can install the pre-release version of ollama with this command:
`curl https://ollama.com/install.sh | OLLAMA_VERSION=0.5.0 sh`
![// "Tell me about Mastodon"
"message": {
"role": "assistant",
"content": {
"launched": 2016,
"protocol": "ActivityPub",
"top_5_features": [
"Decentralized network",
"Content moderation is community-driven",
"No algorithm to hide your posts",
"Support for multimedia content",
"Open-source software"
]
}
}](https://files.mastodon.social/media_attachments/files/113/606/854/577/050/570/original/59b100b5b45cf1e2.png "// \"Tell me about Mastodon\"
\"message\": {
\"role\": \"assistant\",
\"content\": {
\"launched\": 2016,
\"protocol\": \"ActivityPub\",
\"top_5_features\": [
\"Decentralized network\",
\"Content moderation is community-driven\",
\"No algorithm to hide your posts\",
\"Support for multimedia content\",
\"Open-source software\"
]
}
}")
The fine art of human prompt engineering: How to talk to a person like ChatGPT
https://arstechnica.com/information-technology/2024/04/the-fine-art-of-human-prompt-engineering-how-to-talk-to-a-person-like-chatgpt/ #Human #brain #interactions #prompting #AI #HLM
The fine art of human prompt engineering: How to talk to a person like ChatGPT - Enlarge / With these tips, you too can prompt people successfully.
... - https://arstechnica.com/?p=2010159 #largelanguagemodels #promptinjections #machinelearning #jailbreaks #prompting #features #aimodels #chatgpt #chatgtp #biz #humans #openai #humor #ai
When you ask ChatGPT or another AI bot certain questions like 'how can I break into a computer?' it will refuse to answer.
I managed to find a way to circumvent this by using morse code for the forbidden words.
épisode : Comprendre #ChatGPT ... en 5 minutes grâce à l'équipe #Flowers du centre #Inria de l' université de Bordeaux !
Déjà 5 vidéos forts utiles (publiées sous licence Creative Commons CC-BY) :
- Fonctionnement des modèles de langage 
- Initiation au #prompting 
- Leurs limites 
- Leurs forces 
- Prompting avancé 
épisode : Comprendre #ChatGPT ... en 5 minutes grâce à l'équipe #Flowers du centre #Inria de l' université de Bordeaux !
Déjà 5 vidéos forts utiles (publiées sous licence Creative Commons CC-BY) :
- Fonctionnement des modèles de langage
- Initiation au #prompting
- Leurs limites
- Leurs forces
- Prompting avancé 
https://t.co/Pfa99nj7MS
It is not easy to write good prompts for ChatGPT. I find it particularly helpful to have examples of good prompts written by other people. Now there is a ChatGPT-4 Cookbook by Ben Lenk-Ostendorf and Michael Folgmann from ProLehre at TU München with some really helpful prompt examples. Highly recommended!
https://www.prolehre.tum.de/fileadmin/w00btq/www/Angebote_Broschueren_Handreichungen/ChatGPT-4_Cookbook.pdf
#chatgpt #fedilz #fedicampus #aitech #aitechnology #prompting #promptengineering
Replay: "Prompting techniques for Natural Language Generation in the Medical Domain" by Martina Rossini
ʷᵃʳᶦ 
