gzigu (@gzigu@mastodon.social)
Did you know that LLMs are biased based on the language of content? That's because they are extremely good at pattern matching, it's not a fault, it's a feature! The real problem is the data they are being trained on
https://brokenpipe.blog/chatgpt-exposed-beyond-the-magic-for-non-techy-people/
#ChatGPT #gemini #GeminiAI #grokai #openai #machinelearning #artificialintelligence #DeepSeek #censorship #jailbreaking #alibaba #microsoft #google #news #research #technology #tech #blog #blogging
Show Original Post
WebRadio_LesBouffonsDuPeuple (@WebRadio_LesBouffonsDuPeuple@mastodon.social)
Voici en ligne la première émission du 2 septembre 2025 sur les risques psychiques de l'usage immodéré des IA conversationnelles.
#IA #IAconversationnelle #chatGPT #effpndrementPsychique
Show Original Post
psybernaut (@psybernaut@mastodon.social)
Just tried it myself and it goes into meltdown.
https://uk.news.yahoo.com/chatgpt-goes-completely-haywire-ask-100013895.html?guccounter=1
Show Original Post
martin_ueding (@martin_ueding@bonn.social)
📝 Was LLMs über mich zu wissen glauben
Es gibt ja immer mehr Anbieter von LLMs, die man als Chatbot nutzen kann. Und diese werden dann auch zur Hintergrundrecherche genutzt. Da wollte ich mal schauen, was das Teil so über mich zu wissen glaubt.
Ganzer Beitrag: https://martin-ueding.de/posts/was-llms-uber-mich-zu-wissen-glauben/
Show Original Post
Bluedepth (@Bluedepth@mastodon.social)
I’m really digging working with #ChatGPT, the solution it gave me to bring my email archives into #Obsidian is really rocking my world when it comes to raw elegance. The email bodies are markdown files, then all the attachments are stored in a _resources folder, and then it makes an index.md just for kicks! I had a bogus attachment, brought the Python error to the system and it rewrote the code to include detection and skip to gracefully handle the error! Bravo! Really happy! #email #infotech
Show Original Post
habr (@habr@zhub.link)
Анализ системного промпта tool-deep-research
Вот и дошли мои руки покопаться во внутренностях системного промпта tool-deep-research ( ссылка ) Сразу скажу, что возможностями Deep Research я пользовался редко, но в тех моментах, где оно действительно требовалось, он не раз меня выручал. Особенно это касается всяких финансовых вопросов: кредиты, вклады, инвестиции (решения принимаю все еще сам, но понимая конкретную конечную сумму и понятно расписанные условия, принимать решения стало значительно легче) и когда надо быстро погрузиться в новую для себя тему.
https://habr.com/ru/articles/946698/
#джейлбрейк #промптинжиниринг #промпты #промптинг #chatgpt
Show Original Post
ArticleforuWebsite (@ArticleforuWebsite@mastodon.social)
ChatGPT for Customer Support: A Beginner’s Guide #AI #ArtificialIntelligence #Automation #beginnersguide #businesssolutions #Chatbots #ChatGPT #ConversationalAI #customerservice #CustomerSupport #DigitalAssistance #HelpDesk #MachineLearning #OnlineSupport #TechGuide #technology #userexperience #VirtualAssistants
https://articleforu.com/chatgpt-for-customer-support-a-beginners-guide/?fsp_sid=3210
Show Original Post
seeyouindisneyland (@seeyouindisneyland@infosec.exchange)
Content warning:AI makes people dumb. Especially new hires.
Recently™️, I had the pleasure of being the onboarding buddy for one of our newest pentesters at work. Straight outta uni with a shiny new BSc (but weirdly enough they're almost in their 30s - but I ain't HR so I don't care about your CV gaps).
So this somewhat adult person joins the company, gets onboarded, gets their stuff, gets their intro to our IT landscape and whatnot. This person decides to not listen to the very basic intro instructions, i.e. this is how to VPN, that's the way to all necessary info blog posts, here's how you do X and Y.
Person then proceeds to open up ChatGPT right in front of me and asks "How can I do X and Y at $company_name?"
The answer, obviously, is 🐴💩. So I tell that person "Hey, I don't think ChatGPT will solve your problems for our internal things, ya know? I can show you how to do this..." and they go "Nah thanks, I'm good. Got my answer from ChatGPT already"
Chat, are we cooked? 😖
#PenetrationTesting #ITSecurity #Hacking #WhiteHatHacker #NewHires #ChatGPT #WeAreFucked
Show Original Post
ishii00141 (@ishii00141@mastodon-japan.net)
#ChatGPT の計算。一つだけ間違えてる。
8271 × 0.7 = 5789.7
4356 × 0.7 = 3049.2
9724 × 0.7 = 6806.8
5931 × 0.7 = 4151.7
6840 × 0.7 = 4788.0
1283 × 0.7 = 898.1
4759 × 0.7 = 3321.3
2501 × 0.7 = 1750.7
3167 × 0.7 = 2216.9
9042 × 0.7 = 6329.4
Show Original Post
hassmeister (@hassmeister@mas.to)
#ffmpeg gefällt mir immer besser, seit mir #ChatGPT bei den Befehlszeilen hilft. Die sind bisher sauber und funktionieren. :)
Show Original Post
qiita (@qiita@rss-mstdn.studiofreesia.com)
コンテキストエンジニアリングの源流へ、AIと心理学
https://qiita.com/makotosaekit/items/c4d37b91258ec44c0cbf?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items
#qiita #AI #プロンプト #ChatGPT #LLM #コンテキストエンジニアリング
Show Original Post
metadarko (@metadarko@sigmoid.social)
Which model is your favorite?
I keep experimenting, but the real choice… is never easy.
And most of all, how do you use them?
#AI #LLM #OpenSource #Gemma3 #ChatGPT #GPT #Qwen #DeepSeek #AICommunity #Ollama #OpenWebUI #huggingface #API
Show Original Post