Is China Leading the AI Race ?

Good morning.

We’ve been covering the latest AI news every single week for close to a year now.

And while the news out of Silicon Valley is coming at us hard and fast, we must not forget that other countries are also throwing the kitchen sink (read: vast budgets) at the development of AI.

Indeed, when it comes to AGI—artificial general intelligence—or even ASI - artificial superintelligence—the first nation to develop this will have a strategic advantage over all other nations.

There are parallels to be drawn between the Manhattan Project (the race to develop the atom bomb) and research in AI, for sure.

Now there is, of course, discussion about whether AGI or ASI are even achievable; it sure feels like there’s a weapons race going on. Giant amounts of money is being thrown at Artificial intelligence, from silicon chips to investments in research talent (earning up to 1M$ per year). At the end of last week, AI researchers in the US got a stark reminder that they’re facing stiff competition.

Chinese tech firm Sensetime launched an update to its large language model called “SenseNova” that leaves GPT-4 Turbo and LLama3 in the dust across all key benchmarks.

In other news : Chinese AI firm Shengshu introduced a text-to-video model that reportedly rivals OpenAI’s unreleased Sora model in performance and capability.

We usually don’t get a lot of news out of China and the sources are at this point questionable. But … if true there are some people going to be shuffling uneasily on their chairs somewhere in the Bay Area.

Welcome to the Blacklynx Brief.

AI NEWS
The AIPhone ?

• Apple is in talks with OpenAI to potentially include its AI technology in the upcoming iPhone iOS 18 update. These discussions are highlighted as part of Apple's broader AI strategy, set to be showcased at June's WWDC event. Additionally, Apple is considering a collaboration with Google to license Gemini, indicating a significant push to integrate advanced AI features from multiple tech leaders into its platform, aiming to make substantial AI tools accessible to over 1.4 billion iPhone users.

• OpenAI has expanded access to the memory feature for ChatGPT to more Free and Plus users, enhancing its capability to remember details across conversations. This new function allows users to direct ChatGPT to retain specific information for future interactions and review these memories in the settings menu. The introduction of memory aims to streamline user interactions by reducing repetitive inputs and adapting to individual preferences, although its effectiveness will evolve as the feature is refined.

  

  Source : OpenAI

• During the past week, a new AI model known as "gpt2-chatbot" has emerged on Chatbot Arena with advanced capabilities reminiscent of high-end AI models. While it has no official documentation, the model identifies itself as a creation of OpenAI. This development has sparked speculation that OpenAI might be testing a new version of their technology, possibly GPT 4.5, in stealth mode. The model’s impressive performance across various tasks like reasoning and coding has caught the attention of the AI community. A few days later the GPT2 chatbot disappeared from the platform again.

• Google has launched Med-Gemini, an advanced AI tool designed for healthcare that outperforms other top models on most medical tasks. This model uniquely integrates web search data to deliver accurate and current responses to complex health questions. Its capabilities extend to analyzing medical visuals and records, and simulating realistic conversations about patient cases, marking significant progress in AI's application in healthcare.

• GitHub has introduced Copilot Workspace, an AI-powered tool designed to assist developers in planning and initiating projects using natural language within their coding libraries. This new feature integrates with GitHub repositories, allowing developers to outline their goals verbally rather than starting with code. It further enhances productivity by generating preliminary plans and starter code that can be refined and tested directly within the Workspace.

• Eight major U.S. newspapers, including The NY Daily News and The Chicago Tribune, have filed a lawsuit against OpenAI and Microsoft. They accuse the companies of using millions of their articles without permission to train AI models like ChatGPT, which they argue unfairly competes with them as news sources and misattributes information. It does seem that these lawsuits ultimately will end up in OpenAI paying the newspaper for the use of their data and starting a lawsuit might just be the most opportune opening move for these media companies.

Quickfire News

• Apple is poaching AI experts from their competitors, mainly Google, to set up a new AI lab in Switzerland.

• Tesla plans over $10 billion in investment in AI training and inference to enhance competitiveness.

• xAI, backed by Elon Musk, is nearing $6 billion in funding, potentially boosting its valuation to $18 billion.

• Huawei leads a Chinese consortium to develop high-bandwidth AI chips by 2026, aiming to lessen reliance on Western tech.

• Google expands its Virginia and Indiana data centers with a $3 billion investment and launches a $75 million AI Opportunity Fund.

• Elon Musk escalates his legal battle with OpenAI by subpoenaing ex-board member Helen Toner. Musk wants to find out what happened during the board shenanigans at OpenAI.

• Italy is investing $10 billion in its semiconductor industry to become a leading European chip producer.

• OpenAI partners with the Financial Times to use its content for AI training and develop AI features for its readers.

• Baidu finalizes a deal to bring Tesla's Full Self Driving System to China, also accessing Baidu’s mapping data.

• Adobe introduces VideoGigaGAN, enhancing video upscaling by 8 times on its platforms.

• Meta faces criticism over its Advantage Plus AI ad platform due to erratic performance and budget overruns.

• Microsoft reports a strong Q1 2024 revenue boost from AI and cloud computing.

• The U.S. Department of Homeland Security establishes an AI Safety and Security Board for national infrastructure advice.

• Cathy Wood of ARK Invest declares Tesla the most pivotal AI project globally, especially for autonomous robotaxis.

• NIST launches NIST GenAI to test and advance AI technology, following President Biden’s AI Executive Order.

Closing Thoughts

That’s it for us this week.

Clear trends are emerging : not only are vast amounts of money being thrown at the development of artificial intelligence, potentially even more cash is invested in the hardware that will run it all.

A second trend is the move towards mobile AI - with Apple trying to strike a deal with OpenAI for ChatGPT integration into iOS18.

Have a nice weekend.

If you want a version of this newsletter with pictures etc. Be sure to subscribe below:

https://www.blacklynxbrief.com/subscribe