- AI Emergence
- Posts
- Nvidia is the (boastful) darling of the industry!
Nvidia is the (boastful) darling of the industry!
Along with: Excitement building for WWDC!
Analytics Vidhya (Curated by Kunal Jain)
June 06, 2024
Hey there,
NVIDIA stock was 5% up yesterday and it is 32% up in the last month alone! Compare that to AMD (6% up in the month) or Intel (flat) - and you can see the optimism. Every time you think that there are a lot of expectations built into the price - Jensen Huang comes out to surprise you with what they are building!
Now coming back to the Newsletter.
What would be the format? Every week, we will break the newsletter into the following sections:
The Input - All about recent developments in AI
The Tools - Interesting finds and launches
The Algorithm - Resources for learning
The Output - Our reflection
Table of Contents
Nvidia's CEO, Jensen Huang, announced that the company will upgrade its AI chips every year, with the Blackwell Ultra chip coming in 2025 and Rubin in 2026.
At the Computex trade show in Taiwan, Nvidia also revealed new tools and software, highlighting the massive impact of AI.
Huang highlighted Nvidia's success from AI spending and plans to expand its customer base beyond cloud-computing giants to various industries and government agencies.
He discussed the advantages of Nvidia's computing in reducing costs and energy consumption, as well as enhancing AI capabilities in PCs through collaborations with Microsoft and hardware partners.
Additionally, Nvidia is releasing a new server design, the MGX program, and offering free access to its Nvidia Inference Microservices (NIM). (source)
Apple is expected to make significant AI advancements at its upcoming WWDC, focusing on new features in iOS 18 and other operating systems, including audio transcription.
The company plans to process AI data within a "virtual black box" called "Apple Chips in Data Centers" (ACDC), using only Apple's hardware for cloud-based AI processing.
This approach aims to combine the power of cloud processing with the privacy of on-device processing, ensuring that data is secure and inaccessible to Apple or external threats. (source)
Also, The conference is expected to introduce a revamped Siri, featuring advanced voice-activated capabilities and new AI features, primarily for newer iPhone models. (source)
Mistral has launched its first generative AI model for coding, called Codestral. Designed to assist developers with tasks such as completing functions, writing tests, and filling in partial code, Codestral was trained in over 80 programming languages.
Although Mistral describes the model as "open," its license restricts commercial and internal business use, likely due to the inclusion of copyrighted content in its training data.
With 22 billion parameters, Codestral requires significant computing power. Despite some performance advantages, it faces criticism for practical usability and potential issues with code quality. (source)
An open letter titled “Right to Warn,” signed by current and former staff members from OpenAI, Google DeepMind, and Anthropic, cautions against the growing recklessness and secrecy in AI development by major tech firms.
This letter aims to raise public awareness about the risks posed by rapid AI advancements and follows a series of similar warnings from notable figures and organizations.
Building on these concerns, the whistleblowers demanded greater transparency and more protections for those within AI companies. Specifically, they called for:
1. End restrictive nondisparagement and nondisclosure agreements to allow employees to voice concerns freely.
2. Support a culture of open criticism and establish an anonymous reporting process for safety issues.
3. Implement democratic accountability and transparent governance to oversee AI development. (source)
OpenAI LLC is resuming its robotics research after a three-year hiatus. The company had previously shut down its robotics group in July 2021, before the surge of interest in generative AI driven by ChatGPT.
Co-founder Wojciech Zaremba believed in the initial approach but acknowledged that some components were missing for achieving artificial general intelligence (AGI).
The reboot aligns with recent developments where robotics companies have been leveraging ChatGPT and large language models (LLMs) (source)
Recently introduced search feature, AI Overviews, which provides AI-generated answers to certain queries, has increased user satisfaction by providing comprehensive answers to complex questions but came into the talks because of some feedback highlighted odd and erroneous results, with fake screenshots shared on social media.
Google acknowledged these issues, explaining that AI Overviews are designed to provide accurate information by integrating with core search ranking systems and avoiding the typical "hallucinations" of other AI models
Improvements like detecting nonsensical queries, limiting satirical and user-generated content, and refining AI triggers for health and news topics were implemented. (source)
Perplexity has launched Perplexity Pages, allowing users to create and share in-depth articles, reports, and guides.
These pages can include sections, videos, and images, and can be tailored to different audience levels.
Content from Perplexity Pages has started appearing in Google’s AI Overviews and featured snippets. (source)
Stability AI announced the release of Stable Diffusion 3 (SD3) Medium, a 2-billion-parameter AI model, on Hugging Face for June 12th. Designed to run locally on consumer PCs, SD3 Medium enhances photorealism of hands, faces, and typography, and better follows prompts.
While it excels in some areas like typography, it is not flawless. The model is suitable for both consumers and businesses and can be fine-tuned for specific applications.
Stability AI has withheld its most powerful 8-billion-parameter model, potentially for economic reasons. (source)
Tool: Headshot Pro
Problem Statement: Need professional headshots but don't have the time or budget for a photoshoot? HeadshotPro uses AI to generate high-quality, professional headshots quickly and affordably.
How to Access:
Sign up or log in to HeadshotPro.
Upload your photos.
Choose from various professional styles and backgrounds.
Enter your email to receive your generated headshots.
Accept the terms and conditions and start the process.
Receive your professional headshots via email within minutes.
Utilization: Use your AI-generated professional headshots for your LinkedIn profile, company website, or personal branding without the hassle of scheduling a traditional photo shoot.
Get The tool Access: Here
In the recent episode of Leading with Data, I had a chat with Rodolphe Katra. We talk about his journey into Biomedical devices, his views on AI, Robotics, Responsible AI, and Generative AI, and his vision for what’s coming.
Former OpenAI researcher Leopold Ashenbrenner recently appeared on Dwarkesh's podcast, where he discussed several important topics, including CCP espionage, his departure from OpenAI, and the future of AGI. Notably, Ashenbrenner predicts that AGI could be achieved as early as 2027 and emphasizes the importance of understanding these timelines. He has also penned a series of essays titled "Situational Awareness," which delve into AGI timelines in greater detail.
In a recent interview with Prof. Hinton, he talks about how he chooses problems to work on and the power of intuition. Every time I hear Prof. Hinton, my appreciation of his work, his depth on the subject, and his ability to articulate it in simple terms just goes up a notch! Must watch the episode for every AI professional.
I am looking forward to what Apple has in store at WWDC - there is a lot of banking on what it shows and announces for its AI roadmap. In typical Apple style, I expect the use of the word “Artificial Intelligence (AI)” to be about 5 times in the entire Session, but some very creative, complex names are being coined for what would be applications for AI!
In terms of the content - I think the OpenAI deal is coming out along with new potential capabilities of iOS 18. But, the one I am most excited about is how Apple approaches on-device computing and the architecture that would enable AI without being trained on that data directly!
What are you most excited about at WWDC? How many times do you think Apple will mention AI as a word?
Do let me know.
Regards
Kunal Jain
How do you rate this issue of AI Emergence?Would love to hear your thoughts |
Reply