Apple's AI To Run Entirely On-Device?

Plus, Grok gets vision and Meta's AI Might be in Your DMs

In partnership with

Good morning. It’s Monday, April 15th.

Did you know: You can file an extension on your individual tax returns here?

In today’s email:

  • Grok-1.5 Vision

  • Meta’s AI Might be in Your DMs

  • Apple’s iOS 18 AI May Run On-Device

  • 5 New AI Tools

  • Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

In partnership with

85% of all AI Projects Fail, but AE Studio Delivers

If you have a big idea and think AI should be part of it, meet AE.

We’re a development, data science and design studio working with founders and execs on custom software solutions. We turn AI/ML ideas into realities–from chatbots to NLP and more.

Tell us about your visionary concept or work challenge and we’ll make it real. The secret to our success is treating your project as if it were our own startup.

Today’s trending AI news stories

Musk's xAI Debuts Grok-1.5: A Multimodal AI Rivaling GPT-4V and Gemini Pro 1.5

xAI, led by Elon Musk, has introduced Grok-1.5 Vision, its first multimodal AI model, equipped with image understanding and operational capabilities such as flowchart recognition, code writing, and calorie computation from nutritional data. Grok-1.5 targets existing Grok users and select testers for initial testing.

Noteworthy examples demonstrate Grok-1.5's abilities, including converting flowcharts to Python code and determining calorie counts from nutritional information. xAI also introduced the RealWorldQA benchmark, assessing multimodal AI's spatial recognition in real-world scenarios.

Benchmark comparisons reveal Grok-1.5's competitive edge over counterparts like GPT-4V and Gemini Pro 1.5, particularly excelling in RealWorldQA tests. Access to RealWorldQA data is available for download, enhancing transparency. Read more.

Is Meta’s AI in Your DMs?

Meta AI, Meta's AI chatbot, has been introduced on Instagram, enabling users to engage in conversations, seek information, and generate content with text prompts.

This follows Meta's earlier introduction of Meta AI in September 2023 and its subsequent integration into platforms like Facebook Messenger and WhatsApp. While not universally available yet, Meta AI has started appearing in Instagram's search bar, allowing direct interaction within Direct Messaging, which is concerning due to the potential for sensitive messaging being shared with AI servers.

Users can request definitions, story headlines, or images of specific subjects. However, observations suggest Meta AI's functionality remains consistent across Meta's platforms, indicating a standardized integration approach across the ecosystem. Read more.

Apple's First AI Features in iOS 18 Reportedly Won't Use Cloud Servers 

Apple's iOS 18 is set to introduce AI features, with the initial wave expected to operate entirely on-device, bypassing cloud servers.

Bloomberg's Mark Gurman reports that Apple's large language model powering these features won't rely on cloud processing, though some cloud-based AI functions may still be offered, possibly through partnerships with Google, OpenAI, or Baidu.

It's unclear if iOS 18 will integrate Apple's own ChatGPT-like chatbot, with discussions suggesting potential integration with Gemini or other large language models. Future plans may involve Apple's proprietary cloud-based generative AI features, supported by recent server acquisitions. iOS 18 rumors also hint at generative AI enhancements across various apps. Read more.

🖇️ Etcetera

5 new AI-powered tools from around the web

Deblank Colors is an AI-powered palette generator that accelerates design projects. Customizable, interactive tool offers color theory guidance, mockup visualization, and smooth integration for creative workflows.

Evelyn is an open-source AI tutor enhances education with quizzes, mind maps, and flashcards for interactive learning experiences.

LaunchPod is an audio content creation tool for podcasts, audiobooks, and education. Features include AI voice technology, voice cloning, content scheduling, and an AI content builder.

MagicTime is a tool for realistic time-lapse video generation using metamorphic simulators. Features model training, inference, DiT-based architecture integration, extensive documentation, and ChronoMagic dataset. Apache 2.0 licensed with citation encouragement.

Cassidy AI is an AI platform that automates business tasks by integrating with existing tools like Notion, Slack, and LinkedIn.

arXiv is a free online library where researchers share pre-publication papers.

AI Creates Comics

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, apply here.