Jio Will Create New AI Systems Similar to ChatGPT for India, says Ambani

By

-

August 29, 2023

Jio create AI systems similar ChatGPT India Ambani — Image Credits: AGM

Mukesh Ambani, the chairman of Reliance Industries Limited (RIL), claimed that RIL’s Jio would launch new AI systems for India that would be comparable to OpenAI’s ChatGPT. Ambani pledged to make AI accessible to all Indians and to take the lead in developing AI models and solutions that are tailored to India. Ambani announced this at RIL’s 46th Annual General Meeting (AGM).

“Jio promises AI to everyone, everywhere. And we shall deliver,” said Ambani at Reliance’s 46th Annual General Meeting. Additionally, he stated Jio Platforms’ intention to provide AI-driven solutions spanning numerous industries, making sure that the benefits of AI are available to all facets of the Indian population, including people, businesses, and governmental bodies.

This news is noteworthy because it appears to be in response to OpenAI CEO Sam Altman’s remarks that India’s AI efforts are “hopeless,” which he later corrected as having been misinterpreted.

OpenAI CEO Sam Altman spoke at the Economic Times Conversations event in Delhi, where Rajan Anandan, the previous CEO of Google India inquired about his advice on how Indians could develop something similar to ChatGPT in India. Sam responded by adding that while one could attempt to construct something similar in India, it would be “pretty hopeless.”

When the video of the interview went viral, Tech Mahindra CEO CP Gurnani interpreted Altman’s statement as a “challenge”. The CEO of OpenAI later responded to Gurnani’s tweet, claiming that the question he was given was the incorrect one and that his response had been taken out of context. Nevertheless, there have been speculations that Ambani’s announcement is in line with these occurrences.

Jio, which debuted in September 2016 and is now India’s largest mobile operator. Ambani noted during the AGM on Monday that Jio has amassed a customer base of more than 450 million (45 crore) since its launch. Having forayed with such massive success, RIL now wants to tap into the lucrative field of generative AI.

Zepto Raises $200 Million, Becomes First Indian Unicorn in Nearly a Year

By

Suchitra Shenoy

-

August 29, 2023

Zepto becomes first Indian unicorn of 2023

Zepto, the quick commerce aka 10-minute grocery delivery platform, has become the first Indian unicorn (a privately-held startup company valued at over a billion dollars) of 2023. The startup raised $200 million in fresh funding at a valuation of $1.4 billion.

Zepto said it raised funds from US-based investment firms StepStone Group and Goodwater Capital “in the midst of the deepest downturn in capital markets in over a decade.” Existing backers, including Nexus Venture Partners, Glade Brook Capital, and Lachy Groom, also participated in the deal.

Started by two 19-year-old Stanford dropouts in 2021, Zepto promises grocery delivery in 10 minutes. The company competes with SoftBank-funded Swiggy and Blinkit, which are all betting on fast deliveries in the quick commerce sector. However, CEO Aadit Palicha stated that Zepto has the quickest delivery time of 13 minutes.

Zepto plans to use the raised funds to go deeper into the seven Indian metros it currently serves. The company makes a monthly revenue run rate of $50 million – $60 million from Delhi, Mumbai, and Bengaluru alone.

Despite loss-making, a majority of Zepto’s so-called dark stores—densely located warehouses—generate cash, and the company aims to be profitable in 12-15 months. As per Palicha, the company will consider expanding to new cities after a year. Post May 2022, when it last raised funds, Zepto was valued at $900 million.

Qualcomm Unveils Its First ‘Console-First’ Snapdragon G-series Chips for Handheld Gaming Devices

By

Suchitra Shenoy

-

August 29, 2023

Qualcomm releases its first Snapdragon G-series chips

American chipmaker Qualcomm unveiled its Snapdragon G-series line for handheld gaming devices. Currently, the series encompasses G1, G2, G3, and a Snapdragon G3x Gen2 (the most powerful of the lot). According to Qualcomm, the new G-series lineup is designed to meet the performance and feature demands of dedicated gaming devices.

The Razer Edge 5G, currently powered by the Snapdragon G3x Gen1, is getting upgraded to the Snapdragon G3x Gen2. Among the prominent names of handheld console brands, Thundercomm, Huaqin, Inventec, and AyaNeo are some that have joined forces with Qualcomm to launch Snapdragon G-series-powered gaming consoles in the coming days.

Introducing the new #Snapdragon G Series — the Snapdragon G1 Gen 1, G2 Gen 1, G3x Gen 2 gaming platforms and the G3x Gen 2 Handheld Reference Design. Devices powered by these platforms will be the best place to play all your games. 🕹 pic.twitter.com/ByXE5xz0Mt
— Snapdragon (@Snapdragon) August 23, 2023

Mithun Chandrasekhar, senior director of product management at Qualcomm Technologies Inc., stated, “The new generation of Snapdragon G Series powered devices will be the best place for gamers to play their favorite titles, offering them the ability to choose from cloud, console, Android, or PC while on-the-go.”

The Snapdragon G3x Gen2 has a Kyro CPU (8 Core) and Adreno A32 GPU. Its CPU performance is 30% faster than its predecessor’s, and its GPU performance is twice as fast. The Snapdragon G3x Gen2 can power a handheld device with cross-platform gaming capabilities and devices for PC, Android, cloud, and remote console gaming.

SeamlessM4T by Meta is a Multimodal AI Model for Speech and Text Translations

By

Tejaswini Kasture

-

August 28, 2023

Meta's SeamlessM4T, Multimodal AI Model for translating speech and text — Image Credits: AD

Meta launched SeamlessM4T (Massively Multilingual & Multimodal Machine Translation), a cutting-edge AI translation model, both multimodal and multilingual, facilitating seamless communication through speech and text across diverse languages.

According to the SeamlessM4T research paper, Meta developed the SeamlessM4T model by using one million hours of open speech audio data for self-supervised speech representation with w2v-BERT 2.0. They combined this with a multimodal corpus named SeamlessAlign, containing aligned translations covering more than 470,000 hours. By adding human-labeled and pseudo-labeled data totaling 406,000 hours, Meta created a robust dataset. This enabled them to develop the first multitasking and multilingual system, SeamlessM4T.

SeamlessM4T provides a wide range of functions. It recognizes speech in almost 100 languages, translates speech to text for nearly 100 input and output languages, translates speech to speech for nearly 100 input and 36 output languages, converts text to text in nearly 100 languages, and transforms text to speech with support for around 100 input and 36 output languages.

Meta has made a public release of SeamlessM4T under CC BY-NC 4.0, inviting researchers and developers to build on this work. Additionally, they are sharing the metadata of SeamlessAlign, the largest open multimodal translation dataset ever.

In addition, they have further simplified the process for the developer’s community to conduct data mining on their individual monolingual datasets using SONAR, a comprehensive collection of text and speech sentence encoders, as well as stopes, a Meta library for processing multimodal data and parallel data mining. All these advancements are backed by fairseq2, Meta’s advanced sequence modeling library.

Users can explore the seamless communication translation demo through this link. This research demo supports translation for around 100 input and 35 output languages. It allows users to record a sentence, choose upto three languages for translation, and view transcriptions and translations.

India and UAE Sign MoU for IIT Delhi Campus in Abu Dhabi

By

Tejaswini Kasture

-

August 28, 2023

IT Delhi campus in Abu Dhabi — Image Source: siasat

The Ministry of Education, the Indian Institute of Technology Delhi (IIT Delhi), with Abu Dhabi Department of Education and Knowledge (ADEK) have officially signed a MoU to establish the first IIT Delhi campus in Abu Dhabi.

The signing ceremony took place in the presence of UAE President HE Sheikh Mohamed Bin Zayed Al Nahyan and Indian Prime Minister Shri Narendra Modi. The MoU was signed by Professor Rangan Banerjee, Director of the IIT Delhi, Shri Sunjay Sudhir, Indian Ambassador to the UAE, and HE Mubarak Hamad Al Mheiri, ADEK Under Secretary.

IIT Delhi – Abu Dhabi campus is set to commence its academic programs in 2024, featuring a comprehensive degree in bachelor’s, master’s, and Ph.D. programs. It will also establish research centers dedicated to sustainable development studies, as well as computing and data science. The curriculum will encompass various programs, including Artificial Intelligence, Energy and Sustainability, Mathematics, Computer Science and Engineering, and Computing, along with various other fields of sciences, engineering, and humanities.

IIT Delhi, Abu Dhabi, will enhance the academic, research, and innovation landscape in Abu Dhabi by collaborating with prominent institutions such as the Technology Innovation Institute, Khalifa Institute, Mohamed bin Zayed University of Artificial Intelligence, Hub71, and New York University Abu Dhabi. This collaboration will aim to provide complementary programs, pioneer cutting-edge research, and increase the growth of the local startup ecosystem.

Graduates of this campus will become part of a prestigious alumni network of graduates from 23 campuses throughout India. IIT Delhi has recently ranked among the world’s top 50 institutions in engineering and technology. It has also achieved a top 30 ranking for global employability in the 2022 QS World University rankings.

Expressing his views on social media, Shri Dharmendra Pradhan, the Indian Union Minister of Education, Skill Development & Entrepreneurship, highlighted the significance of this MoU signing. He emphasized that the establishment of the IIT Delhi campus in Abu Dhabi, in the presence of the Prime Minister of India, unfolds a new chapter in the internationalization of India’s education. This campus is a testament to India’s innovation and expertise, symbolizing the strong friendship between India and UAE.

This marks a significant stride in our educational internationalisation and is testament to India’s innovation prowess. Education is the bond that unites us, it's the spark that ignites innovation. Together, we will leverage this power for mutual prosperity and global betterment. https://t.co/TFfmSFWzsQ
— Narendra Modi (@narendramodi) July 15, 2023

India’s AI-skilled Workforce Expanded 14 Times in 8 Years, says LinkedIn Report

By

Sahil Pawar

-

August 28, 2023

According to a LinkedIn report, the number of workers in India with artificial intelligence skills surged 14 times between January 2016 and June 2023. India now ranks among the top five nations for the growth of AI talent, with Singapore, Finland, Ireland, and Canada.

The ‘Future of Work Report‘ from professional network platform LinkedIn underlined the growing importance of AI in the workplace.

In India, according to the report’s findings, 43 percent of members reported seeing an increase in the use of AI at work during the previous year. 71 percent of Gen Z professionals in the nation now acknowledge that learning AI skills could improve their employment prospects as a result of the spike.

According to the report, AI and Machine Learning are among the digital skills that every two out of every three workers stated they will learn this year. About 57 percent of CEOs want to increase the usage of AI in their organizations in 2024. This demonstrates that companies are actively retraining their current employees in AI skills to stay flexible.

Additionally, 91 percent of senior executives in India, exceeding the global average of 72 percent, acknowledged the rising importance of AI skills. According to the survey, seven out of ten professionals felt that having soft skills like creativity and problem-solving enabled them to approach their work from a new angle.

According to Ashutosh Gupta, Country Manager of LinkedIn India, as AI defines the future of work, India is realizing the value of human potential and the crucial role that AI and soft skills will play in creating a workforce of the highest caliber possible.

Hugging Face Raises $235 Million in Series D Funding Round

By

Sahil Pawar

-

August 28, 2023

Hugging Face, an open-source repository for AI models, has recently raised $235 million in a Series D investment round involving Google, Amazon, Nvidia, and Salesforce as demand for AI model access rises.

Hugging Face is one of several companies gaining from the interest of investors in generative AI. Its most recent round leaves the company at a $4.5 billion valuation. Hugging Face acts like GitHub for machine learning and other AI models, codes, and datasets.

Google, Amazon, AMD, Intel, IBM, Nvidia, and Salesforce were among the investors in this round. All these companies have made large investments in the foundation models for generative AI or the processors that operate these models.

The latest amount of funding will be used, according to Hugging Face CEO Clement Delangue, to grow the team, which is currently at 170, and invest in more open source AI and platform building. The investment, according to him, also highlights how crucial the open-source community is to the growth of the generative AI community.

Delangue said, “It’s a validation not only for Hugging Face, but also for the larger open-source AI ecosystem. Companies want to develop AI internally, and this requires open-source AI.”

Numerous well-known generative AI models are available on Hugging Face, including Stability AI’s Stable Diffusion and Meta’s Llama 2 and its variants. Also, in February this year, Hugging Face chose Amazon Web Services (AWS) as its preferred cloud provider, which explains its participation in the funding.

Alibaba Introduces AI Models Qwen-VL and Qwen-VL-Chat

By

Sahil Pawar

-

August 28, 2023

Alibaba on Friday unveiled new artificial intelligence models that, according to the company, are capable of understanding images and having conversations that are more complicated than those of the company’s prior products. The announcement comes as the competition for technological leadership across the world intensifies.

The Chinese technology behemoth announced that its two new models, Qwen-VL and Qwen-VL-Chat, will be open source, allowing researchers, academia, and businesses around the world to utilize them to develop their own AI apps without having to invest time and money in training their own systems.

Alibaba claimed that Qwen-VL could produce captions for photographs and react to open-ended questions about various images. In contrast, Qwen-VL-Chat, supports more “complex conversation,” such as comparing several image inputs and responding to series of questions. Alibaba claims that Qwen-VL-Chat is capable of composing stories, producing images from user-inputted photos, and solving mathematical problems represented as pictures, among other things.

Alibaba used the input of a Chinese-language hospital sign as an example. By analyzing the sign’s image, the AI can respond to queries regarding the location of specific medical departments.

The two most recent models from Alibaba are based on the Tongyi Qianwen large language model that was released earlier this year. Similar to Qwen-VL-Chat, the most recent version of OpenAI’s large language model GPT-4 can comprehend visuals and answer in text.

This month, the startup with its headquarters in Hangzhou made two more AI models, Qwen-7B and Qwen-7B-Chat, open source. Alibaba will not receive any licensing fees from the open-source distribution, but it will gain additional customers for its AI model at a time when its cloud division is trying to resume growth as it gets ready to go public.

AI Cannot Replace Human Intelligence in Court Rulings, says Delhi HC

By

Sahil Pawar

-

August 28, 2023

The Delhi High Court has ruled that human intelligence or the human component in the adjudicative process cannot be replaced by artificial Intelligence. It declared that ChatGPT cannot serve as the foundation for making legal or factual assessments in a court of law.

An effort to use ChatGPT findings to support a party’s claim of trademark infringement was declined by Justice Prathiba M. Singh. The HC emphasized that there is still uncertainty surrounding the accuracy and dependability of AI-generated data, and that such a tool is only suitable for basic research or analysis.

“The aforementioned tool cannot serve as the foundation for a court of law’s decision on any legal or factual matters,” said Singh.

According to her, the response of a Large Language Model-based chatbot like ChatGPT, which is sought to be relied upon by counsel for the plaintiff, depends on a host of factors, including the nature and structure of the user’s query, the training data, etc. Furthermore, there is a possibility that AI chatbots will produce inaccurate responses, fictitious precedents, and inventive facts, the HC said.

The court made this observation while it processed a case involving a trademark infringement claim brought by a luxury brand against a partnership company that manufactured and sold shoes. The plaintiff’s attorney claimed that a shoe name was its trademark in India and presented comments from ChatGPT regarding its reputation in court.

Meta Introduced Code Llama, an AI-powered Tool for Coding and Debugging

By

Tejaswini Kasture

-

August 28, 2023

Meta launches Code Llama for Coding and Debugging — Image Source: Llama

Meta has unveiled Code Llama, an advanced Large Language Model (LLM) that possesses the ability to develop code as well as natural language descriptions related to code in response to prompts given in either code or natural language.

Currently, Code Llama can be used for both research and commercial purposes without any cost. It is built on top of Llama 2 and is offered in three versions: Code Llama – the core code model, Core Llama – Python, designed specifically for Python language, and Code Llama – Instruct, which is fine-tuned to understand natural language instructions.

There are three variants of Code Llama, featuring 7B, 13 B, and 34B parameters, respectively. Each model is trained with a dataset of 500B tokens containing code and code-related information. The Code Llama – Python is fine-tuned using a dataset containing 100B tokens of Python code.

In addition to its code completion features, it can also be used for debugging purposes. It supports popular programming languages, including Java, Python, Typescript (Javascript), Bash, C++, Bash, and PHP.

Beyond its necessity for generating extended programs, having a longer input sequence brings new possibilities for a code language model. For example, users can give the model more information from their existing code or pass the entire code to make its generated code more relevant. These features can be helpful to developers while debugging when trying to fix issues in large codebases.

Developers can request access to Llama from the Meta Webpage, and their training methodologies are also accessible through the GitHub repository.

Overall, Code Llama is designed to cater to the needs of software engineers across diverse sectors. This includes professionals working in research, industry open-source projects, and businesses.