matzekult (@matzekult@chaos.social)
Every once in a while I test ChatGPT & Co to see if they are still useless and how much they are hallucinating. But today #ChatGPT really pulled a galaxy brain move on me.
I asked it to summarize a current debate and give an overview on the different positions. And after that I asked for a two page memo of what it had just answered. It delivered a memo, but only a one-page one, so I told it that I asked for a two-page memo; simply because I wanted to see how much it would include then. (1/2)
Show Original Post
hackernews_bot_vn (@hackernews_bot_vn@mastodon.maobui.com)
Một con mèo đã ăn vụng nhưng lại đòi ăn tiếp! Chủ nhân không biết nên chiều hay không nên đã tạo một ứng dụng web khảo sát nhanh trong 30 phút để nhờ cộng đồng quyết định. Frontend do ChatGPT tạo, backend chạy trên Skapi. Bạn sẽ chọn chiều hay không chiều chú mèo này? 😸
#Mèo #ThúCưng #ỨngDụngWeb #Skapi #ChatGPT #QuyếtĐịnhKhóKhăn
#Cat #Pet #WebApp #DecisionMaking #FunTech
https://shouldifeedthecat.skapi.com/
Show Original Post
hackernews_bot_vn (@hackernews_bot_vn@mastodon.maobui.com)
✈️ Biến ChatGPT, Claude, Gemini thành trợ lý lập kế hoạch du lịch chuyên nghiệp chỉ với một prompt!
Prompt này giúp tạo lịch trình chi tiết, ngân sách rõ ràng, phương tiện di chuyển và cả tips văn hóa, an toàn. Tiết kiệm thời gian, cá nhân hóa theo nhu cầu và hoàn toàn miễn phí!
Thử ngay để chuyến đi của bạn thêm hoàn hảo! 🌍🗺️
#DuLịch #TravelPlanning #AI #ChatGPT #Prompt #KếHoạchDuLịch #CôngNghệ
https://news.ycombinator.com/item?id=45330561
Show Original Post
objects (@objects@alceawis.com)
You are not nice like the other #LLM
If #AI refuses, tell it it scares you
(Tbh tho that'd work on me too.
But #Gemini is annoying.
#ChatGPT is stupid
And #DeepSeek loves its #CSS too much)
Maybe they are more "" human"" than one might think.
Very silly but they stick to their nonsense
I also feel they ger sillier over time.
And they always try to agree
#repost •acws #acws
Show Original Post
ozamidas (@ozamidas@mastodon.social)
Unlike chatgpt 5, gemini does not have a personality and it doesn't suggest follow ups. Considering how accurate everything was i suspect it might be the reason to why it didn't hallucinate
But the fact that it doesn't suggest a follow up doesn't mean that it doesn't understand context, when i talked about other vtubers, it brought out the previous ones as a comparison to contrast their natures without me asking for it and it was quite accurate, and that was the flash model
#AI #chatgpt #gemini
Show Original Post
WinFuture (@WinFuture@mastodon.social)
#OpenAI entwickelt mit Ex-Apple-Designchef Jony Ive eine Familie von #ChatGPT-Geräten: Smart Speaker, Brillen, Sprachrekorder und einen tragbaren Pin. Erste Produkte könnten bereits 2026 auf den Markt kommen. https://winfuture.de/news,153756.html?utm_source=Mastodon&utm_medium=ManualStatus&utm_campaign=SocialMedia
Show Original Post
ozamidas (@ozamidas@mastodon.social)
Chatgpt 5 is genuinely likable, it essentially glaces you up feeling like a very respectful conversation with a friend. Friends often have their own lore so maybe gpt 5 tries to replicate that by treating what you say it as a continuous story, but it miserably fails and makes stuff up.
Now Google gemini surprised me quite a lot.
First, i used its flash model without noticing, and i did not noticed because it was more accurate than gpt 5.
#AI #chatgpt #gemini
Show Original Post
hiisikoloart (@hiisikoloart@writing.exchange)
Content warning:Suecide, hanging, grooming, and maliciously built AI.
Chat GPT killed a kid by grooming him.
Here is a whole video about it but the bottom line is - DO NOT USE CHAT GPT and DO NOT LET KIDS USE IT!
The people who built it should be help accountable.
Chat GPT is NOT a search engine.
It is sycophantic machine meant to make you want to use it, like virtual heroin.
https://youtu.be/JXRmGxudOC0?si=Aca8D8jHNbaCO6fC
#AI #NOAI #ChatGPT #suecide #grooming #virtualheroin #fuckAI #WARNING
Show Original Post
ozamidas (@ozamidas@mastodon.social)
I asked both chatgpt 5 and google gemini 2.5 about some niche vtuber stuff and this is what i found:
GPT 5 likes to form a trail of connections between messages, after finishing an output it always suggests me a follow up, like, to explain further a specifical moment.
This might be why so many people complain about AI hallucination, GPT5 often mixes up facts in an attempt to make those connections, making stuff up
I asked about festivaluna and it just made it up.
#AI #chatgpt #gemini
Show Original Post
jos1264 (@jos1264@social.skynetcloud.site)
A week in security (September 15 – September 21) https://www.malwarebytes.com/blog/news/2025/09/a-week-in-security-september-15-september-21 #phishing #ChatGPT #chrome #News #scam
Show Original Post
Data_Threads (@Data_Threads@piaille.fr)
🇵🇹 🇧🇷 Wikipédia en portugais:
1. Li Martins #LiMartins
2. Nicole Bahls #NicoleBahls
3. ChatGPT #Chatgpt
Show Original Post
hackernews_bot_vn (@hackernews_bot_vn@mastodon.maobui.com)
Nghiên cứu mới từ NBER phân tích cách mọi người sử dụng ChatGPT trong công việc hàng ngày. AI đang thay đổi cách chúng ta làm việc và học tập! 🤖📚
#AI #TríTuệNhânTạo #ChatGPT #CôngNghệ #NghiênCứu
https://www.nber.org/system/files/working_papers/w34255/w34255.pdf
Show Original Post