AI: Weekly Summary. RTZ #569

1. Google Gemini 2.0 here: Google answered OpenAI’s 12 days of AI Shipmas, video LLM Sora et al, with a bagful of LLM AI releases, all at one go. This included developer focused Gemini 2.0 driven tools and services. Early reception has generally been positive, with developers intrigued by Gemini’s Mariner agentic capabilities, and Astra, its visual AI identification systems. More details on the Google line up available here, including a new Trillium AI chip that delivers higher processing speeds for Gemini 2.0. Bloomberg summarizes it here. Google like OpenAI, is focused on setting the roadmap towards autonomous reasoning and agentic capabilities for its next gen Gemini 2.0 AI systems. More here and here.

   

2. Apple Intelligence here with iOS 18.2: Apple’s long awaited Apple Intelligence integration with OpenAI’s ChatGPT is here, along with a plethora of other features, as part of iOS 18.2. Early reviews are generally constructive, more or less. The highlight feature for now is Siri’s integration with ChatGPT from OpenAI. Apple Visual Intelligence with Google and others echoes similar features being rolled out separately by Google, OpenAI, and others. Apple continues to focus on its trust and privacy prioritized AI features, built bottom up and inside out for the multi-billion users of its local devices, applications, and services ecosystem. More here.

   

3. Microsoft weighs in on AGI: Microsoft AI head Mustafa Suleyman had a detailed conversation with the Verge on a range of AI issues including the timing for AGI. The whole podcast is worth a weekend listen with the key items highlighted here. Of particular interest is his timeframe for AGI being two to five years of the upcoming hardware and software generations from the LLM AI companies, and AI GPU/infrastructure companies like Nvidia. This is a bit in contrast to shorter term time frames from OpenAI, Anthropic and others. The debate over the timing and substance of AGI continues. More here.

   

4. AI Power beyond Nuclear: Traditional power sources may have a role to play in scaling Power for the upcoming multi gigawatt AI Data Centers, if Exxon Mobil’s plans are realized. They’re outlining plans for 5 plus gigawatt natural gas powered options for AI data centers, with built in carbon capture capabilities. Those would rival some of the options being pursued by big tech companies with various nuclear power utilities. To date, Microsoft, Amazon, Google and others have all indicated interest in nuclear power options. Traditional natural gas options, along with hydro and other power sources would be an additional way to accelerate the power requirements for a trillion plus dollar makeover of the global AI data center industry. More here.

   

5. AI Worlds beyond Video: LLM AI models are rapidly evolving to generate video and interactive worlds. OpenAI’s broad release of text to video Sora is the latest example, with generally positive reviews and anticipation. Competition is also fierce with companies and AI technologies from China. Google also announced its Genie 2 models, designed to ramp up interactive worlds from images. And startups like World Labs are also focused on generating 3D interactive worlds from images. The applications for these technologies are of course immediately apparent in Hollywood and video game markets. But the potential applications beyond may surprise in the long run. More here.

   

Other AI Readings for weekend:

1. AI Eating Browsers, Google, OpenAI, Apple et al. More here.

   

2. AI & tech investment boom in Asia. More here and here.

   

Up next, the Sunday ‘The Bigger Picture’ tomorrow. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)