Welcome to Get The Gist, where every weekday we share an easy-to-read summary of the latest and greatest developments in AI—news, innovations, and trends—all delivered in under 5 minutes! ⏱

In today’s edition:

• Google Launches New AI Learning Companion Tool

• Amazon Develops Its Own Custom AI Chips

• Google DeepMind Open-Sources AlphaFold 3

• Baidu Launches AI-Powered Smart Glasses

• And more AI news….

Top Developments

Google

Google Launches New AI Learning Companion Tool

The Gist: Google’s “Learn About” AI tool, powered by the LearnLM model, provides educational-style responses that go beyond typical chatbot answers, adding context, visuals, and interactive learning elements.

Key Details:

• Unlike traditional chatbots, Learn About uses educational formatting with features like vocabulary builders and “why it matters” sections to deepen understanding.

• Responses are visual and interactive, using educational sources and offering related topics for further exploration.

• The tool successfully interprets even tricky questions, showing promise for refining AI’s role in learning and answering “common misconceptions.”

Amazon

Amazon Develops Custom AI Chips to Reduce NVIDIA Dependence

The Gist: Amazon is advancing its in-house AI chips to decrease reliance on NVIDIA’s powerful GPUs, leveraging its custom processors to cut costs and boost AI capabilities for cloud and AI partner applications.

Key Details:

• Amazon’s Annapurna Labs, which previously developed the Graviton and Trainium processors, is now creating specialized AI chips, expected to be detailed next month.

• The chips are part of Amazon’s long-term strategy to lower costs in data centers and reduce dependency on NVIDIA, whose GPUs dominate the AI market but are costly and supply-constrained.

• Amazon’s primary AI partner, Anthropic, is using these custom processors to power its Claude AI model, directly competing with NVIDIA-backed AI systems.

• Amazon joins other tech giants, like Alphabet and Meta, in developing proprietary AI chips, with Meta recently launching its own MTIA chips and Google releasing its new TPU, Trillium, to improve AI training speeds.

Google DeepMind

Google DeepMind Open-Sources AlphaFold 3

The Gist: Google DeepMind has open-sourced AlphaFold 3, granting scientists worldwide access to its protein prediction tools, which are poised to drive advances in drug discovery and molecular research.

Key Details:

• AlphaFold 3’s model code and training weights are now accessible for academic, non-commercial research, making cutting-edge protein prediction widely available.

• Released in May 2024, AlphaFold 3 offers 50% improved accuracy over previous versions, predicting the structure of proteins, DNA, RNA, and other molecules.

• The model supports Google DeepMind’s partnerships with Eli Lilly and Novartis to accelerate AI-driven drug development.

• DeepMind recently used AlphaFold to predict over 200 million protein structures, sparking breakthroughs in targeting complex diseases.

Quick Gist

Fastino launched from stealth with $7 million in funding to develop task-optimized enterprise AI models that run on CPUs, avoiding the high costs of GPU infrastructure (Read More).

Microsoft partnered with Iconem to create a detailed digital twin of St. Peter's Basilica using AI and photogrammetry, enhancing virtual accessibility and preservation of the iconic landmark (Read More).

Microsoft has launched Copilot in Power BI Mobile apps, enabling users to generate report summaries and insights for data-driven decisions on the go (Read More).

OpenAI's new AI model, Orion, reportedly shows only modest improvements over GPT-4 despite early training successes, as the company shifts its strategy to address data limitations and enhance model performance post-training (Read More).

Apple is reportedly developing a smart home camera integrated with Apple Intelligence and health-focused AirPods, both set for mass production in 2026 (Read More).

Anthropic's chief Dario Amodei announced plans to release Claude 3.5 Opus, addressing delays due to extensive computational requirements and a focus on enhanced capabilities (Read More).

Google's AI service, Gemini, is set to introduce new features allowing users to upload and modify files, with updates expected in the coming weeks (Read More).

Baidu launches AI-powered smart glasses in China, aiming to compete with Meta and Snap in the AI wearables market (Read More).

THAT’S IT FOR TODAY!