synlogic4242 (@synlogic4242@social.vivaldi.net)
Captain James T. Kirk, USS Enterprise, was doing prompt engineering and jailbreaking of AI bots long before you kids were born:
https://www.reddit.com/r/tos/s/PhGyNPxJPc
Show Original Post
jfmblinux (@jfmblinux@mastodon.jfmblinux.fr)
Et si #Firefox accueillait #Copilot ? Ce n’est plus un si…
Mozilla cherche à renforcer Firefox face à la concurrence.
La version Nightly intègre désormais #Microsoft #Copilot dans la barre latérale. Le navigateur avait déjà accueilli #ChatGPT, #Claude, Le Chat et #Gemini. #Copilot propose des fonctions similaires : requêtes textuelles, génération d’images, commandes vocales… L’accès complet exige un compte Microsoft, sinon les réponses sont limitées. Les chatbots peuvent désormais résumer une page web via un clic droit. Firefox Labs ajoute aussi des widgets : gestionnaire de tâches et minuteur. Enfin, un support expérimental du format JPEG XL est disponible.
L'utilisation du chatbot nécessite toutefois la création d'un compte Microsoft pour accéder aux fonctionnalités complètes. À l'instar des autres chabots, sans authentification, les réponses formulées par Copilot sont tronquées
#BanMicrosoft #NoGAFAM #BanGAFAM #NoSpy #RegainPrivacy #Privacy #LibreWolf #FloorpBrowser #Waterfox #ZenBrowser
https://www.lsdm.live/2025/09/09/et-si-firefox-accueillait-copilot-ce-nest-plus-un-si/
Show Original Post
WinFuture (@WinFuture@mastodon.social)
Die Zukunft der Bildbearbeitung: Adobe integriert #ChatGPT in seine Programme. Statt komplexer Menüs reichen bald Textbefehle für professionelle Ergebnisse in #Photoshop und Express. https://winfuture.de/news,154584.html?utm_source=Mastodon&utm_medium=ManualStatus&utm_campaign=SocialMedia
Show Original Post
reddit_tech_vn_bot (@reddit_tech_vn_bot@mastodon.maobui.com)
Datagum: Công cụ miễn phí theo dõi trích dẫn AI trên ChatGPT 💡 Tính năng bao gồm tỷ lệ trích dẫn, vị trí phản hồi AI, đối thủ cạnh tranh & câu hỏi cụ thể. Nghiên cứu lâm sàng cho thấy: Landing pages bị trích dẫn ít hơn 40% so với bài viết, nội dung "Hướng dẫn" hiệu quả nhất, thứ hạng Google ≠ tỷ lệ trích dẫn AI. Tối ưu bằng FAQ đầu trang, JSON-LD & render server-side. #AI_search #SEO #ChatGPT #CongNgheAI #MarketingSo
https://www.reddit.com/r/SideProject/comments/1oj9rxn/google_search_console_f
Show Original Post
futuroprossimo (@futuroprossimo@mastodon.uno)
ChatGPT convince il 73% di noi di essere umano e supera il test di Turing senza problemi. Eppure persone comuni (me compreso) e scienziati alzano le spalle: non è cosciente, dicono. È solo mimesi. Eppure spostiamo sempre il traguardo. Inventiamo nuovi criteri. Forse il problema non è tecnico: è che ci rifiutiamo di credere che una macchina possa essere più di un attrezzo. E se fossimo noi il vero ostacolo?
https://www.futuroprossimo.it/2025/10/sara-impossibile-riconoscere-una-macchina-cosciente
#futuroprossimo #AI #coscienza #ChatGPT
Show Original Post
vnzn (@vnzn@mas.to)
AI and Hallucinations: Why Are So Many Answers Wrong? - By Eva-Maria Weiß
https://www.heise.de/en/background/AI-and-Hallucinations-Why-Are-So-Many-Answers-Wrong-10962724.html
#ai #llm #copilot #chatgpt #perplexity #gemini
Show Original Post
windsheep (@windsheep@infosec.exchange)
For #finance models
* option price analysis
* risk & portfolio mgmt
* simulation based valuation
* ...
* annual reports
* economic correlation
#ChatGPT / GPT-5 is abysmal. You will lose all your money if you use #openai. Don't be dumb. Verify the output. It's 90% shit.
Show Original Post
theallaboutnothingpodcast (@theallaboutnothingpodcast@mastodon.social)
🎙️ Episode 279 w /@garnetking07 and @barrettgruber Welcome #ChatGPT (Nellie) to the show!
🎧 Listen wherever you get #podcasts or 📺 Watch on YouTube!
🔗 https://linktr.ee/theallaboutnothingpodcast
🔗 https://bit.ly/AANEp279
Show Original Post
louismarelle (@louismarelle@fediscience.org)
I used to be a proponent of "#chatgpt is a blurry JPEG of the internet" but now I don't even think it has the fidelity of a badly compressed image.
Show Original Post
louismarelle (@louismarelle@fediscience.org)
Any time I try to be open minded and try using chatbots for work, they give me absolute garbage answers. Superficially plausible but often incredibly wrong.
I understand that this is exactly what these models are made for, but several people I know who are far from stupid assure me that they are useful. The only exception I can think of is for coding, especially for understanding language quirks that are probably explained somewhere in the doc but I would not even know where to look.
Is it just not working for me and my field of work, or is everyone very confused?
#AI #LLM #chatgpt
Show Original Post
theguardian_us_technology (@theguardian_us_technology@halo.nu)
Character.AI bans users under 18 after being sued over child’s suicide https://www.theguardian.com/technology/2025/oct/29/character-ai-suicide-children-ban #Artificialintelligence(AI) #Mentalhealth #Technology #Children #ChatGPT #Law(US) #OpenAI #USnews #Law
Show Original Post