Поиск по этому блогу

Search1

123

среда, 26 июля 2023 г.

Quietly Selfish with My Apple

Can't read or see images? View this email in a browser
 

Apple loves to build an ecosystem for developers. Recently, with the launch of Vision Pro, Apple announced tools such as Xcode, SwiftUI, ARKit, and TestFlight. That’s not all, they also came up with Reality Composer Pro for developers to create immersive experiences with “unbounded 3D content”. Looks like Apple has a clear vision for developers, the only limitation being, it’s for ‘Apple developers only’.

https://media3.giphy.com/media/v1.Y2lkPTc5MGI3NjExMmNiNGdjOTR0OTRhdTUwc2kwbHBiYThvMmpxMW16dWdycHZkdTRyciZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/ZFEcY7LpPXqbJ6kzui/giphy.gif

The tech giant has replicated the same closed-door mindset when it comes to developing generative AI technology. Despite not opening up its technology to others, Apple has never shied away from using the open-source ecosystem. For instance, Apple’s Ajax system is built on top of Google Jax, the search giant’s machine learning framework, and Apple’s system runs on Google Cloud, which the company uses to power cloud services, alongside its own infrastructure and Amazon’s AWS.


The tech community abhors this approach. Aleksa Gordic, ex-Google DeepMind/ Microsoft ML engineer vented on Twitter. “I wish the Llama 2 licence was somehow applicable to fundamental AI research to force Apple to publish their internal research. They never give anything in return, and yet benefit enormously from research coming from other big-tech companies. It just doesn’t feel right.” 


Compared to Apple, other big techs are contributing heavily to the development of AI technology. According to Hugging Face, Meta makes the highest contribution to the open-source community, followed by Google and Microsoft.


Industry experts say that the current AI development is going through a crucial phase and it is imperative that Apple drops its walled-garden policy for the sake of AI. 


Read the full story here.




Brin is Back


Google cofounder Sergey Brin is back in business. The trusted old-timer is helping the tech giant in its war against Microsoft-funded OpenAI. Brin has been reportedly working on AI project Gemini, a ChatGPT rival and a brainchild of DeepMind’s Demis Hassabis. 


In January this year, Brin filed his first request in years for access to the code of Gemini. Brin aims to bring back trailblazing AI innovation to the company after the LaMDA fiasco in the summer of 2022 acted as a major setback when a senior software engineer claimed the company’s AI chatbot LaMDA was sentient. 


Read more here.




Moderators Fight Back

https://media1.giphy.com/media/v1.Y2lkPTc5MGI3NjExNWNqOTkxdDk5d3psdG96bnRsY216bzV0Mm9uNjdwd2oxenFtNGk3OCZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/UtVBBd2b50TEHxCRmL/giphy.gif

Stack Overflow is in an extremely bad state. Data shows that the question-and-answer website for programmers has lost about 50% of its traffic over the past year and a half. It has also experienced a sharp decline in the number of queries on the platform. This came after the launch of ChatGPT, which has drawn a huge chunk of the programmers who have been thronging the platform to solve their technical queries.


Having witnessed the misery of the platform, the moderators have now buckled up and are working to boost the quality of the platform. Since ChatGPT and other LLM-based chatbots are prone to hallucinations, a sharp focus on providing only correct answers on Stack Overflow can be a solution to bring users back to the platform.


Read more here.




Papers Meta Presented at ICML 2023


Meta AI presented some groundbreaking papers at the 40th International Conference on Machine Learning (ICML) in Honolulu, Hawaii. Their research covers diverse areas, including domain adaptation with label shift using Efficient Label Shift Adaptation (ELSA), learnability of episodic reinforcement learning in Reward-Mixing Markov decision processes (RMMDPs) with EM2 algorithm, the versatile Masked Trajectory Models (MTM) for prediction and control, and Hyperbolic Image-Text Representations with MERU to organise visual and written ideas hierarchically. 


Additionally, they introduced Hiera, a simplified and faster Hierarchical Vision Transformer for image and video recognition tasks. These papers showcase state-of-the-art advancements in AI and ML.


Read more here.

     

TAUSIF ALAM & AMIT RAJA NAIK

Wednesday, Jul 26, 2023 | Was this email forwarded to you? Sign up here

     
   

DOWNLOAD OUR MOBILE APP

Stay Connected

info@analyticsindiamag.com

© 2023 Analytics India Magazine

   
Facebook
Twitter
LinkedIn
Youtube
Instagram
   
 
Analytics India Magazine | 280, 2nd floor, 5th Main, 15 A cross, Sector 6, HSR layout Bengaluru, Karnataka 560102

Комментариев нет:

Отправить комментарий

Примечание. Отправлять комментарии могут только участники этого блога.