2 Minute Weekly AI - Week 1

Author

Axel Flerén
January 05, 2025

Happy New Year, now kicking it off with the first Weekly AI newsletter of 2025!

The following is a ~2-minute read featuring News & Insights in AI for Week 1, 2025.

—

Updates & News - Week 1

NVIDIA Files Patent for AR Glasses

NVIDIA’s latest patent introduces an approach to AR glasses, replacing traditional backlighting with digital holography. By leveraging self-interference with incoherent light, the system enhances optical occlusion while reducing power consumption and the overall size compared to conventional AR displays. This subtractive method selectively darkens only the areas necessary for AR, keeping the rest of the user’s vision clear.

The system uses ambient light rather than generating its own, achieving superior energy efficiency. The patent covers advanced optical management, including polarization control and interference optimization, along with a neural framework for real-time distortion correction and adaptive display adjustments. NVIDIA’s innovation could redefine AR with faster, more efficient, and responsive experiences.

AR Glasses will most likely be one of the big future technology innovations, possibly replacing smartwatches and phones.

Check out these already made AR Glasses from Meta, to get a feel of how AR Glasses work:

AI Companies Slashing Prices

The generative AI market is entering a fierce price war as companies compete for dominance. Alibaba Cloud has led the charge, cutting prices by up to 85%, with its visual language model Qwen-VL seeing the steepest reductions. This trend reflects the growing rivalry among Chinese tech giants such as Tencent, Baidu, and ByteDance, all battling for market share with minimal product differentiation.

In the West, OpenAI and Google have responded by offering cheaper, scaled-down models. Meanwhile, Anthropic has taken a different approach, increasing prices for its Haiku model while introducing the more affordable Sonnet 3.5.

With open-source models like Meta’s Llama gaining traction, AI companies are balancing premium pricing with aggressive cuts to stay competitive and secure long-term growth.

OpenAI Confirms Plans to Become a For-Profit Company

OpenAI has officially announced plans to shift from its nonprofit-controlled structure to a for-profit model by 2025. The reorganization aims to attract more funding beyond its historic $6.6 billion raise, transforming OpenAI into a Delaware Public Benefit Corporation (PBC). This move will allow traditional equity investments while retaining its mission to serve the public good.

However, OpenAI’s pivot has drawn criticism. Former executives, like Miles Brundage, warn of governance ambiguities, suggesting the nonprofit arm may grant excessive control to the PBC. Similarly, Jan Leike, now with Anthropic, criticized the move as uninspired, arguing it compromises OpenAI’s lofty AGI mission for narrower philanthropic efforts in healthcare and education.

While the PBC model aims to balance financial goals with social impact, critics fear profit-driven motives may undermine long-term alignment and equitable AGI benefits.

Will be interesting to see how this all plays out in the end with the lawsuit from Elon Musk against this transition. Musk has now also received support from various AI experts and advocacy groups as he challenges the company's profit-driven direction

Setbacks for OpenAI and xAI with Delays

OpenAI’s Media Manager, designed to give creators control over their content in AI training, has yet to launch despite its 2025 promise. Initially promoted as a tool to remove copyrighted material, its delay raises questions about compliance burdens and whether it addresses deeper intellectual property issues. Legal experts remain skeptical, especially as lawsuits pile up, casting doubt on its effectiveness in mitigating risks.

Meanwhile, xAI’s Grok 3 missed its 2024 deadline, adding to a growing list of delayed AI models. Grok 3, expected to rival GPT-4 and Gemini, promised improved image and conversational capabilities but may now be replaced by an interim Grok 2.5.

While Elon Musk’s ambitious timelines often seem aspirational, the repeated delays highlight the limits of brute-force computing and traditional training methods.

—

That was everything for Week 1, hoping that you found it valuable!

If you want more updates in AI then make sure to follow me on my socials 👇

And if you work in e-commerce and want to avoid wasting valuable human time on manual customer support, book a demo below to find out if an AI customer service employee, available on your website and working 24/7, could help.

(30-day trial available)

Back soon,

Axel