🕶️ Meta's Glasses now with Multimodal AI

Also: Adobe's new AI model

Welcome, AI enthusiasts

Meta and Ray-Ban are expanding their smart glasses collection together with new multimodal AI features. Adobe has released their new generative model, Firefly Image 3, that will directly be integrated into Adobe Photoshop. Its promising better and more realistic results. Also, Microsoft introduces Phi-3-mini, a compact language model that uses optimized training data to achieve the performance of much larger models like GPT-3.5. Let's dive in! 

In today’s insights:

  • Meta's Glasses now with Multimodal AI

  • Adobe adds AI tools to Photoshop powered by updated Firefly Image 3

  • Microsoft Unleashes Phi-3

Read time: 4 minutes

🗞️ LATEST DEVELOPMENTS

Source: Meta

Evolving AI: The Ray-Ban Meta smart glasses are finally here with Multimodal AI features, merging iconic style with cutting-edge technology.

Key Points:

  • Newly released Ray-Ban Meta smart glasses with expanded styles and advanced Meta AI features.

  • Introducing video calling features compatible with WhatsApp and Messenger for a hands-free experience.

  • Continuous improvements and updates, bringing real-time visual translation and interaction through Meta AI.

Details:

In partnership with EssilorLuxottica, the second-generation Ray-Ban Meta smart glasses combine traditional aesthetics with modern technology, ensuring a fit for every face. They offer a range of new styles and lenses, including the vintage-inspired Skyler and a low bridge fit for the Headliner frames. These glasses integrate video calling and AI functionalities, allowing users to interact with their environment and digital content seamlessly.

Why It Matters:

As AI hardware devices are evolving, Meta’s smart glasses are an interesting product in this field as they blend increasingly sophisticated AI capabilities with everyday practicality and style. The Ray-Ban Meta smart glasses are not only trying enhance social interactions with features like live video sharing but also pave the way for real-time AI assistance in our daily visual and auditory experiences.

Below is a video of all the new multimodal AI features for Meta’s glasses.

TOGETHER WITH MAILSPLASH
🔥 Supercharge your email marketing with AI


Evolving AI: Slash your email marketing costs and time dramatically with AI.


With Mailsplash, you will:

  • Save on costs that typically range from $250 to $1000 per email; Mailsplash does it in seconds for just a few dollars per email.

  • Instantly generate high-quality email campaigns and flows, ready to send quickly after integration with your ESP.

  • Focus on creating quality emails, while Mailsplash’s AI collects all the necessary information from your website and products

🌟 Sign up for beta access to get exclusive pricing.

Join Mailsplash today, and start creating high-converting emails in minutes!

Source: Adobe

Evolving AI: Adobe's new generative model, Firefly Image 3, promises more realistic results, launching at Max London.

Key Points:

  • Adobe introduces Firefly Image 3, enhancing realism and detail.

  • Comparison to previous models shows marked improvements.

  • Ethical training practices and new generative tools launched.

Details:

Adobe's latest model, Firefly Image 3, part of its broader Firefly suite, claims to produce more realistic images than its earlier versions. This upgrade includes better understanding of complex prompts, advanced lighting, and text generation. Notably, it excels in depicting dense crowds and a variety of human expressions. During its Max London conference, Adobe also introduced precision editing tools like Generate Background and Enhance Detail that utilize this new model. Additionally, Adobe continues its efforts to use AI ethically by compensating stock contributors and moderating the AI-generated content it trains its models on.

Our Thoughts:

With Firefly Image 3, Adobe isn't just improving technology but also its stance on ethical AI use. It's a noteworthy advance, potentially setting a new standard for realism in AI-generated art and addressing ongoing concerns around data practices. Could this be the model that changes the landscape of digital creativity?

Source: beebom

Evolving AI: Microsoft has officially launched the Phi-3 family, including a variant with up to 128,000 token context length.

Key Points:

  • Introducing Phi-3-mini and upcoming models in the Phi-3 series.

  • Advanced capabilities in small language models

  • Strategic availability on multiple platforms

Details:

Microsoft Research unveiled the Phi-3-mini, a streamlined language model that matches the capabilities of larger models like GPT-3.5. The key to Phi-3-mini's performance is its high-quality training dataset, which consists of efficiently optimized, filtered web and synthetic data. Despite having just 3.8 billion parameters, the Phi-3-mini achieves a 69 percent success rate on benchmarks like MMLU (Massive Multitask Language Understanding) and is capable of operating on smartphones with a minimum of 1.8 GB of memory, as reported by Microsoft. The model features a context window of up to 128K tokens. The family will expand soon, enhancing flexibility and performance across different scales.

Our Thoughts:

More and more big AI companies are releasing smaller AI models as they’re all gearing up to develop and release on-device AI models for our smart phones and devices. Microsoft’s focus on knowledge and reasoning capabilities for Phi-3, combined with web access, could enable more efficient and cost-effective AI models that Microsoft needs for its products and services.

🔎 AI in research

Source: Profluent | The physical structure of OpenCRISPR-1, a gene editor created by AI technology from Profluent

Profluent, an AI-first company, launched OpenCRISPR-1, the first AI-designed, open-source gene editor for editing human DNA, to speed up the development of treatments for genetic diseases. You can read their blog here.

Key takeaways:

  • The company used AI to create many unique CRISPR-like proteins, potentially leading to more effective and adaptable gene editing tools.

  • OpenCRISPR-1 is available for free to researchers, aiming to spread advanced gene editing technologies widely.

  • The breakthrough suggests a future where AI-designed tools could quickly tailor treatments to individual genetic disorders, although immediate healthcare impacts are limited by high preclinical costs.

Why This Matters:

The results point to a future where AI precisely designs what is needed to create a range of bespoke cures for diseases. Medicing combined with AI is the thing that could catapult us into a new age - let’s see how this field develops.

💡 Tip of the Day

With the release of Adobe’s Firefly Image 3, this means that Photoshop will receive numerous updates, as Adobe is directly integrating this new model into Photoshop. Get up to speed with all the changes by watching this detailed video below.

🚀 Create stunning emails in minutes and unleash the strengths of AI-powered copywriting and seamless design customization for your email marketing strategy (typically worth $250 to $1000 per email blast).

Join Mailsplash today to get exclusive pricings. 🎁

🎯 SNAPSHOTS

Direct links to relevant AI articles.

👀 Amazon: Aims to host companies’ custom AI models.

🤖 OpenAI: OpenAI shrugs off Meta’s Llama 3 ascent with new enterprise AI features.

📈 Trending AI Tools

  • 👨🏽‍💼 Sana - AI assistant for work (link)

  • 📝 Jamie AI - AI notetaker for meeting notes (link)

  • 🦉 Trademark Owl - AI trademark registration tool to minimize complications (link)

  • ❤️ Docus AI - Your personal AI health assistant (link)

  • 🔎 Glimpse - AI powered tool designed to streamline online research (link)

  • 🤝 Venturefy - A tool to verify corporate proof to increase trust with customers (link)

Join the conversation

or to participate.