‍

Hello Niuralogists!

‍

Step into this week's edition as we navigate the dynamic realm of artificial intelligence to present you with the most recent breakthroughs. Our primary focus is to dissect the implications of these updates on various aspects of our lives, ranging from workplaces and businesses to policies and individual experiences. In this edition, we'll uncover exciting developments, including OpenAI's delay in releasing ChatGPT's Voice Mode and a Meta alumnus unveiling an AI model simulating 500 million years of evolution.

‍

For deeper insights, continue reading…

‍

OpenAI Postpones ChatGPT Voice Mode Release

OpenAI has announced a delay in the launch of ChatGPT's much-anticipated Voice Mode feature, which was initially expected to roll out within a few weeks of the Spring Update in May. The company cited the need for additional safety tests and refinements as the reason for the postponement. An Alpha test group that was planned for late June has been pushed back by a month to allow for these safety checks. The initial demo of the Voice Mode, which featured a model named "Sky" that bore a resemblance to actress Scarlett Johansson, stirred some controversy, though OpenAI denied any wrongdoing. In the meantime, OpenAI has released a ChatGPT desktop app for Mac users, offering broader integration across files, screen-sharing, and conversations. While the delay is disappointing, particularly after missing the previous launch timeline, it is considered preferable to releasing an underdeveloped product amidst growing competition from companies like Anthropic.

‍

Meta Alum Unveils AI Model Simulating 500 Million Years of Evolution

Meta alum Shubham Sharma launched EvolutionaryScale, an AI research lab dedicated to making biology programmable. The lab, founded by former Meta engineers from the company's protein-folding team, introduced ESM3, a generative language model that simulates 500 million years of evolution to design novel proteins. ESM3, which can follow prompts and generate new proteins, created a green fluorescent protein (esmGFP) in tests. The model's ability to simulate rapid evolution marks a significant advancement, as it achieved in a short time what would naturally take hundreds of millions of years. EvolutionaryScale also announced a $142 million seed funding round led by notable investors, including Nat Friedman and Daniel Gross. The smallest model of ESM3 has been open-sourced, while larger versions are available for commercial use through partnerships with Nvidia and AWS. This development highlights the potential of AI in addressing major challenges like climate change and disease, with pharmaceutical companies standing to benefit the most from novel medicine development.

‍

‍

Apple and Meta Enter Discussions on AI Collaboration

Apple is reportedly in discussions with Meta to integrate the social media giant’s generative AI models into Apple Intelligence, indicating a potential thaw in their previously icy relationship. Apple is looking to expand its AI partnerships beyond OpenAI, considering collaborations with Meta, Perplexity, and Anthropic to join ChatGPT in its new system. This development is surprising given the long-standing rivalry between Apple and Meta, particularly over privacy issues and App Store policies. The rumored deal would allow Meta and other AI partners to offer premium subscriptions through Apple devices, with Apple taking a revenue cut. Apple software chief Craig Federighi hinted at the company's plan to offer multiple AI options during WWDC. This potential partnership could signify a major shift for the rivals, as the AI revolution might bring them together. For Apple, securing a variety of models provides users with the best possible choices, while Meta gains a significant new distribution channel for its AI efforts.

‍

Alibaba Cloud Unveils English Version of AI Model Hub

Alibaba Cloud has expanded its AI offerings globally by launching an English version of ModelScope, its open-source AI model community. This move aims to bring generative AI capabilities to a broader audience of businesses and developers worldwide. Since its launch in China in 2022, ModelScope has become the country’s largest AI model community with over five million developer users. The international version provides access to more than 5,000 advanced AI models and welcomes user-contributed models, fostering a collaborative ecosystem. Unveiled at the 2024 CVPR Conference in Seattle, ModelScope offers a comprehensive suite of tools and resources, including high-quality datasets and customizable modules for model inference, training, and evaluation. Alibaba Cloud's strategy to globalize its AI platform is underscored by its presence at CVPR, where the company demonstrated the capabilities of its Qwen model series and highlighted its commitment to advancing AI research and development.

‍

‍

Ilya Sutskever Launches New AI Initiative

Ilya Sutskever has launched a new AI venture, Safe Superintelligence (SSI), focused exclusively on developing safe and powerful superintelligence, just a month after his official departure from OpenAI. Co-founded with former Y Combinator partner Daniel Gross and ex-OpenAI engineer Daniel Levy, SSI aims to create beneficial superintelligent AI, which Sutskever predicts could emerge within a decade. Unlike OpenAI's shift towards commercial products, SSI will be a pure research organization insulated from short-term business pressures. While funding details have not been disclosed, Gross indicated that raising capital will not be a challenge. This move comes after months of speculation following Sutskever’s board drama and fallout with OpenAI, and SSI promises to build the safety-focused AI lab Sutskever envisioned, attracting ample funding and talent.

‍