آیا روزی خواهد بود که این چت‌بات‌ها و language model ها، همدلی و فهمیدن و گوش کردن فعالانه رو هم بتونن انجام بدن؟

یه باگ امنیتی فاجعه هم پیدا شده

Introducing acropalypse: a serious privacy vulnerability in the Google Pixel's inbuilt screenshot editing tool, Markup, enabling partial recovery of the original, unedited image data of a cropped and/or redacted screenshot.

https://twitter.com/ItsSimonTime/status/1636857478263750656?t=TQmS-nOyUPGeygR3lrbVIA&s=19

و ویندوز:

holy FUCK.

Windows Snipping Tool is vulnerable to Acropalypse too.

An entirely unrelated codebase.

The same exploit noscript works with minor changes (the pixel format is RGBA not RGB)

Tested myself on Windows 11

https://twitter.com/David3141593/status/1638222624084951040?t=zyHKT0W2_xa_306kYr34Qw&s=19

The Sparks of AGI have been Ignited

“In this paper, we report on our investigation of an early version of GPT-4, when it was still in active development by OpenAI. We contend that (this early version of) GPT-4 is part of a new cohort of LLMs (along with ChatGPT and Google's PaLM for example) that exhibit more general intelligence than previous AI models. We discuss the rising capabilities and implications of these models. We demonstrate that, beyond its mastery of language, GPT-4 can solve novel and difficult tasks that span mathematics, coding, vision, medicine, law, psychology and more, without needing any special prompting. Moreover, in all of these tasks, GPT-4's performance is strikingly close to human-level performance, and often vastly surpasses prior models such as ChatGPT. Given the breadth and depth of GPT-4's capabilities, we believe that it could reasonably be viewed as an early (yet still incomplete) version of an artificial general intelligence (AGI) system.”

Paper: Sparks of Artificial General Intelligence: Early experiments with GPT-4

کتاب همین حوالی دوردست رو می‌خونم. دوست داشتم دوستی هم باشه که براش بخونم:

"Writing is saying to no one and to everyone the things that is not possible to say to someone. Or rather writing is saying to the no one who may eventually be the reader those things one has no someone to whom to say them. Matters that are so subtle, so personal, so obscure, that I ordinarily can't imagine saying them to the people to whom I'm closest. Every once in a while I try to save them aloud and find that what returns to mush in my mouth or falls short of their ears can be written down for total strangers. Said to total strangers in the silence of writing that is recuperated and heard in the solitude of reading."

اگه پیام دادید و جواب ندادم در این روزها، اولا که عذرخواهی می‌کنم. دوما که جواب خواهم داد. در حال حاضر جواب دادن به پیام‌ها ازم انرژی زیادی می‌خواد و تواناییش رو ندارم چون ظرفیت ذهنیم فعلا محدودتر از گذشته است و مجبور شدم یه سری چیزا رو محدود یا حذف کنم. ممنون که متوجه میشید. و به زودی جواب میدم

پریروز مدل ModelScope ارائه شد که مشابه مدل‌های dalle2 و stablediffusion عه ولی برای تبدیل متن به ویدیو. یعنی یه متن بهش می‌دید و براتون فیلم درست می‌کنه:

https://huggingface.co/spaces/damo-vilab/modelscope-text-to-video-synthesis

این مثال یه خروجی عه که خودم ازش گرفتم.
متنش هست:
a llama in a college class

این هم مدل جالب جدید دیگه‌ای عه که بهش عکس می‌دی و با stablediffusion سعی می‌کنه نسخه‌های دیگه‌ای ازش رو بسازه:

https://huggingface.co/spaces/lambdalabs/stable-diffusion-image-variations