Google Teases GPT-4o Rival Before I/O

OpenAI’s Lead in AI Voice Assistants May Last Just One Day

What you need to know

AI News Today, OpenAI launched GPT-4o, an advanced AI model capable of interacting with users through audio, vision, and text. However, Google seems poised to counter with its own innovation at Google I/O 2024, starting tomorrow. In a recent teaser, Google previewed an unreleased multimodal AI interface, hinting at major announcements related to AI, Google Search, and other groundbreaking technologies. Stay tuned for exciting updates!
Today, OpenAI unveiled GPT-4o, a new AI model capable of interacting with users through audio, vision, and text. This strategic launch came just one day before Google’s anticipated AI announcements at the Google I/O 2024 developer conference. However, OpenAI’s lead in multimodal AI might be short-lived.
In a teaser posted on X (formerly Twitter), Google previewed an unreleased multimodal AI interface running on an Android phone. The teaser video showcased an AI that visually resembles the Pixel Camera app. It demonstrated the AI’s ability to answer questions about its surroundings by identifying the Google I/O stage and providing details about the upcoming event. While the results are impressive, the pre-recorded nature of the video means they should be viewed with some caution.
Google has hinted at major announcements related to AI, Google Search, and more, set to be revealed at their I/O 2024 event, with the main keynote scheduled for 10 a.m. PT. We will cover all developments live.
Additionally, Google’s Gemini, previously seen as a voice assistant and a chatbot capable of processing images and screenshots, now appears to support vision as an input method. The AI interface can use a device’s camera to answer questions and provide details about the user’s surroundings, similar to the capabilities seen in the Humane AI Pin and Rabbit R1 standalone devices.
However, several details remain unclear, such as whether this functionality will be integrated into Gemini or a different application. The demo suggests that a more advanced version of Gemini could potentially replace Google Assistant, though this remains speculative.
We will learn more during Google’s keynote, but for now, the teaser confirms that a competitor to GPT-4o is on the horizon. Stay tuned for more updates from Google I/O 2024.SOURCE 

Apple Updates Logic Pro for Mac & iPad with AI Features

AI News Apple has launched an updated version of Logic Pro, bringing in all the new features previewed last week. This update comes alongside new versions of both Logic Pro and Final Cut Pro, announced with the latest iPad Pro models. The updates are now available for both Mac and iPad users.

New AI Tools

AI News The latest update to Logic Pro introduces a suite of enhanced AI tools that join existing features like Smart Tempo and the Pitch Correction plug-in, significantly augmenting your music creation process. These new tools leverage advanced AI technology to streamline and enhance various aspects of music production:

• Smart Tempo: This feature automatically manages tempo adjustments, ensuring seamless timing and synchronization across your tracks.
• Pitch Correction Plug-in: Provides precise pitch correction, enabling vocal and instrumental tracks to achieve perfect intonation.

New AI Enhancements:

• AI-Powered Session Players: Including a Bass Player and Keyboard Player, these virtual musicians use AI to follow chord progressions from the Global Chord Track, creating realistic and responsive performances that adapt to your direction.
• ChromaGlow: An advanced saturation plug-in with five distinct styles that emulate the warmth and character of vintage analog hardware, adding depth and richness to your tracks. This tool requires an M1 chip or later.
• Stem Splitter: This powerful tool separates stereo audio files into individual stems for vocals, drums, bass, and other parts, making it easier to apply effects, add new elements, and adjust mixes. This feature also requires an M1 chip or later.

Session Players

The latest update to Logic Pro introduces advanced AI-driven Session Players, including a new Bass Player and Keyboard Player, which join the existing Drummer feature. These virtual musicians enhance your music production by providing realistic, responsive performances that adapt to your compositions. Here’s how they work:

• AI-Driven Performance: The Bass Player and Keyboard Player use advanced AI algorithms to follow chord progressions dictated by the Global Chord Track. This ensures that their performances are harmonically accurate and musically coherent.
• Seamless Integration: These Session Players integrate seamlessly with your existing projects, automatically adapting to the chord changes and musical nuances of your tracks.
• Versatile Applications: Whether you’re composing a new piece, arranging existing music, or experimenting with different styles, these AI musicians provide a versatile and reliable foundation, allowing you to focus on creativity and expression.
• Realistic Sound: The AI models are designed to emulate the playing styles and tonal qualities of real musicians, providing a natural and authentic sound that enhances the overall quality of your music.

ChromaGlow

The latest update to Logic Pro introduces ChromaGlow, an advanced saturation plug-in designed to emulate the warmth and character of vintage analog hardware. This powerful tool is perfect for adding depth, presence, and richness to your tracks. Here’s a detailed look at ChromaGlow:

• Five Distinct Saturation Styles: ChromaGlow offers five unique saturation models, each carefully crafted to simulate the sound of classic analog equipment. These styles provide a range of tonal enhancements, from subtle warmth to intense coloration.
• Versatile Application: Whether you’re working on vocals, instruments, or entire mixes, ChromaGlow allows you to dial in the perfect tone, adding a professional-quality finish to your projects.
• Enhancing Presence and Depth: The saturation effects not only add warmth but also enhance the presence and depth of your tracks, making them sound fuller and more engaging.
• User-Friendly Interface: ChromaGlow features an intuitive interface that makes it easy to select and adjust saturation styles, allowing you to quickly achieve the desired effect.
• System Requirements: This plug-in requires an M1 chip or later, ensuring optimal performance and integration with the latest Apple hardware.

Stem Splitter

The latest update to Logic Pro includes the Stem Splitter, a powerful feature designed to enhance your audio editing capabilities. Stem Splitter allows users to separate a stereo audio file into individual stems for vocals, drums, bass, and other parts. Here’s how it can benefit your music production:

• Detailed Audio Separation: Stem Splitter accurately divides a stereo track into its constituent elements, such as vocals, drums, bass, and other instrumental parts. This precise separation enables greater control over each component of your mix.
• Enhanced Mixing Flexibility: By isolating individual stems, you can apply effects and adjustments to specific parts of a track without affecting the entire mix. This makes it easier to fine-tune your audio and achieve a professional-quality sound.
• Creative Remixing: Stem Splitter opens up new possibilities for creative remixing. You can rearrange, modify, or replace individual elements of a track, giving you the freedom to experiment and innovate with your music.
• Efficient Workflow: The feature is designed to integrate seamlessly into your existing workflow. It simplifies the process of manipulating complex audio files, saving you time and effort.
• System Requirements: Stem Splitter requires an M1 chip or later, ensuring optimal performance and leveraging the advanced processing power of the latest Apple hardware.

Sound Library Updates

AI News The latest update to Logic Pro brings significant enhancements to the Sound Library, providing users with a rich array of new instruments and features designed to inspire creativity and improve workflow. Here are the key updates: Studio Bass: Access six deeply-sampled acoustic and electric basses, offering a wide range of tones and styles to suit any genre. These high-quality samples provide realistic and expressive bass sounds that can elevate your tracks. Studio Piano: Perform with three meticulously-sampled pianos, each capturing the nuanced sound and feel of grand pianos. These pianos offer a dynamic range and detailed resonance, perfect for adding a touch of elegance to your compositions. Chord Tag Loops: This feature allows loops that contain chord tags to automatically populate the chord track when added to a project. This automation simplifies the process of building harmonic structures and ensures seamless integration of loops into your compositions. Producer Packs: Discover new producer packs from renowned artists such as Hardwell, The Kount, and Cory Wong. These packs include a variety of loops, samples, and presets that can inspire new musical ideas and add professional-quality sounds to your projects. Demo Song: Explore the original multi-track project of “Swing!” by Ellie Dixon, available as an in-app demo song. This project serves as an excellent learning tool, allowing you to see how professional tracks are constructed and mixed.

Spatial Audio Enhancements

The latest update to Logic Pro introduces significant enhancements to spatial audio capabilities, providing users with greater flexibility and control over their audio mixes. Here are the key improvements:

• Custom Mixing: Logic Pro now offers downmix and trim options for non-Atmos channel configurations. This feature allows users to create custom mixes tailored to various playback environments, ensuring optimal sound quality across different audio setups.
• Expanded ADM BWF Files: The support for ADM BWF files has been expanded beyond Dolby Atmos. Users can now include settings for stereo and other multi-channel formats within these files. This enhancement broadens the applicability of ADM BWF files, making it easier to manage and export complex audio projects for diverse output requirements.

General Enhancements

The latest update to Logic Pro introduces several general enhancements designed to streamline your workflow and boost productivity. Here are the key improvements:

• Bounce in Place: This feature now includes automatic real-time recording for External Instrument regions or tracks utilizing Logic’s I/O plug-in. This enhancement simplifies the process of capturing live performances and external audio sources, making it easier to integrate them into your projects seamlessly.
• MIDI Routing: Logic Pro now allows MIDI generated by supported software instruments and effects to be routed to the input of other tracks. This capability enables creative layering and complex MIDI routing setups, providing greater flexibility in your music production.
• Editing Efficiency: New key commands have been added for moving, extending, and resizing marquee selections. These shortcuts enhance your editing efficiency, allowing for quicker and more precise adjustments to your audio and MIDI regions.SOURCE 

Announcing GPT-4o: Revolutionizing Human-Computer Interaction

AI News OpenAI is excited to announce the launch of GPT-4o, our new flagship model that takes human-computer interaction to the next level. This innovative AI model can process and reason across audio, vision, and text in real-time, making interactions more natural and efficient.

Key Features of GPT-4o

Multimodal Capabilities:

AI News GPT-4o (“o” for “omni”) is designed to accept and generate outputs in any combination of text, audio, image, and video. It can respond to audio inputs in as little as 232 milliseconds, with an average response time of 320 milliseconds—comparable to human conversation speed. It matches the performance of GPT-4 Turbo on text in English and coding tasks, and it excels significantly in non-English languages, while also being faster and 50% cheaper in the API. GPT-4o particularly shines in vision and audio understanding, outperforming previous models.

Integrated Voice Mode:

Before GPT-4o, Voice Mode in ChatGPT used a pipeline of three separate models, leading to latencies of 2.8 seconds with GPT-3.5 and 5.4 seconds with GPT-4. This process limited the model’s ability to fully capture tone, multiple speakers, and background noises, and it couldn’t output expressive elements like laughter or singing. GPT-4o overcomes these limitations by integrating all modalities into a single, end-to-end model, processing all inputs and outputs through the same neural network. AI News

Performance and Evaluations

Benchmark Achievements:

GPT-4o sets new benchmarks in various areas:

• Text, Reasoning, and Coding: Achieves GPT-4 Turbo-level performance.
• Multilingual, Audio, and Vision: Sets new high watermarks with significant improvements.
• Reasoning: Scores an impressive 88.7% on the 0-shot COT MMLU (general knowledge questions) and 87.2% on the traditional 5-shot no-CoT MMLU.

Audio and Vision:

• ASR Performance: Dramatically improves speech recognition over Whisper-v3, especially for low-resource languages.
• Audio Translation: Sets new state-of-the-art performance on speech translation, outperforming Whisper-v3 on the MLS benchmark.
• Multilingual and Vision Evaluation (M3Exam): Excels in this benchmark, which includes multiple-choice questions with figures and diagrams from various countries’ standardized tests.
• Vision Understanding: Achieves state-of-the-art results on visual perception benchmarks like MMMU, MathVista, and ChartQA, all evaluated with 0-shot CoT.

Language Tokenization:

GPT-4o’s new tokenizer efficiently compresses data across 20 representative languages from different language families.

Safety and Limitations

Built-In Safety:

GPT-4o incorporates safety features across all modalities. Techniques include filtering training data and refining the model’s behavior through post-training. Evaluated using OpenAI’s Preparedness Framework, GPT-4o does not score above Medium risk in cybersecurity, CBRN, persuasion, and model autonomy. This assessment involved comprehensive automated and human evaluations during the training process.

External Red Teaming:

GPT-4o has undergone extensive testing with over 70 external experts in fields like social psychology, bias, fairness, and misinformation. This red teaming helps identify and mitigate risks associated with the new modalities.

Audio Modality Risks:

Acknowledging the novel risks of audio modalities, the initial public release includes text and image inputs and text outputs. Further updates will address the technical infrastructure, usability via post-training, and safety measures necessary for broader modality support. Initially, audio outputs will be limited to preset voices and will adhere to existing safety policies.

Availability and Future Plans

Model Rollout:

GPT-4o is our latest advancement in deep learning, emphasizing practical usability. After extensive efficiency improvements, GPT-4o is available to a broader audience. The rollout begins today with text and image capabilities in ChatGPT’s free tier and higher message limits for Plus users. A new Voice Mode with GPT-4o will be available in alpha within ChatGPT Plus in the coming weeks.

Developer Access:

Developers can access GPT-4o through the API as a text and vision model. GPT-4o is twice as fast, half the price, and offers five times the rate limits compared to GPT-4 Turbo. Support for new audio and video capabilities will be launched to a small group of trusted partners in the API in the coming weeks.
OpenAI invites feedback to identify tasks where GPT-4 Turbo may still outperform GPT-4o, as we continue to improve the model and push the boundaries of AI capabilities.SOURCE 

Conclusion

The advancements in AI News are rapidly evolving, with OpenAI’s GPT-4o setting new benchmarks in multimodal interactions and Google poised to unveil its competitor at Google I/O 2024. GPT-4o, with its ability to process and reason across audio, vision, and text in real-time, promises to revolutionize human-computer interactions by making them more natural and efficient. Its improved performance, especially in non-English languages, audio understanding, and vision capabilities, marks a significant step forward in AI technology.
The updated Logic Pro from Apple also showcases the integration of AI News to enhance music production, providing users with new tools and features to elevate their creative processes.
As these technological advancements unfold, it’s clear that AI is becoming more integrated into various aspects of our digital lives, offering unprecedented opportunities for innovation and efficiency.
For more updates on https://www.arcotgroup.com/ai-innovations-virtual-einstein-lectures-stable-artisan-launch-and-openais-google-challenger/AI News and technological advancements, contact Arcot Group at ArcotGroup.com. Stay informed and ahead of the curve with the latest in AI developments.