Warm greetings and welcome to the second issue of the JC AI Newsletter. The world of artificial intelligence continues to be a thrilling journey of discovery and innovation. As we conclude the second half of July, we've had some major breakthroughs and developments that are set to redefine the landscape of AI going forward.
Specifically, the announcement from Meta about the launch of their Llama2 model as an open-source project has become the talk of the tech town. Available in 7B, 13B, and an astounding 70B parameters versions, both pre-trained and fine-tuned, this model is causing ripples across the AI community. It is emerging as a potent competitor to OpenAI's GPT-4. With a license that authorizes commercial use, the implications for AI applications are nothing short of exciting.
As we dance to the groove of this thrilling news, let's continue to explore what the junction of Investment and AI can create. Stay tuned, stay synced, and let's dive into this exciting new rhythm together!
OP-ED: The AI Chessboard: Voluntary Commitments, Game Theory & International Geopolitics
Let's take a stroll down the complex lane of regulating Artificial Intelligence (AI). Earlier this month, the Biden-Harris Administration managed to secure voluntary commitments from seven leading AI Titans. Sounds promising, right? Yet when you peer through the lens of game theory, things get a tad more complicated. Check out our latest blog post for a closer look into this news.
Jazz Bytes: July 16-31, 2023 AI Digest
The second half of July was bustling with noteworthy news and declarations. Here are the top highlights, complete with links for you to delve deeper!!
- Meta (Facebook) announced the release of the Llama-v2 model as an open-sourced project, both pretrained and fine-tuned models are available with 7B, 13B and 70B parameters along with a commercial use license. They are also available on Microsoft Azure and Amazon AWS, and Hugging Face. (Meta's Llama-2 Website, Llama-2 Paper).
- In a proactive move to manage the risks associated with Artificial Intelligence (AI), the Biden-Harris Administration has obtained voluntary commitments from seven AI industry leaders - Amazon, Anthropic, Google, Inflection AI, Facebook, Microsoft, and OpenAI. These commitments are centered on enhancing the safety, security, and transparency of AI development.
- Cerebras and G42 are partnering to address the ever-growing need for computational power spurred by Generative AI. Last week, they announced two more US-based 4 exaFLOPS AI supercomputers, with the aim of building the world's largest AI supercomputers of 36 exaFLOPS of AI compute by the end of 2024. Read more here
- On the AI product pricing front, Microsoft has released the anticipated pricing of its Microsoft 365 AI-powered Co-Pilot at $30/month per user on top of the regular 365 subscription.
- Google Cloud's revenues reached $8 billion in Q2 2023, with more than 70% of Gen AI unicorns run on GCP.
- A new research paper from Stanford and UC Berkeley, authored by Matei Zaharia and James Zho, studies the changes and degradation of the “performance and behavior” of OpenAI’s ChatGPT LLMs over the last few months.
- Meta and Qualcomm are partnering to run big A.I. models on phones, enabling new large language models, like Llama 2, to run on Qualcomm chips on phones and PCs starting in 2024.
- Snowflake entered the Generative AI game when it unveiled new capabilities to enable developers to build new LLM-enabled applications. This involves new containerization capabilities in Snowpark and a partnership with NVIDIA.
- The Impact of GenAI on K-12 Education
- How AI is Reimagining the Future of Gaming
- Generative AI and the Telecommunication Industry
- Choosing the Right Large Language Model (LLM): Trade-offs between open-source vs closed-box LLMs
- Can AMD Surpass Nvidia's Dominance in the GPU Arena?