Google DeepMind unveils AI model for living cells

Google DeepMind recently announced the release of AlphaFold 3, an advanced AI model designed to enhance our understanding of biological structures and their interactions within cells. This third iteration, originally initiated in 2018, now includes the ability to analyze not just proteins, but also the genetic materials DNA and RNA, and other molecules like ligands which are crucial for signaling and disease processes. The model, developed in collaboration with Isomorphic Labs, aims to significantly speed up the discovery and hypothesis testing of new drug molecules, leveraging AI to predict complex biological structures rapidly compared to traditional experimental methods.

AlphaFold 3 has shown a marked improvement in predictive accuracy over its predecessors and other existing tools, which could revolutionize biological research and drug discovery. It enables scientists to formulate and verify hypotheses at an atomic level within seconds, a process that would traditionally take months or years. This efficiency could be particularly transformative in developing drugs, understanding plant biology, and enhancing food security. DeepMind has also made this powerful tool available to academic and non-commercial users for free, potentially broadening its impact across various scientific fields.

Despite its advancements, AlphaFold 3 faces challenges such as dealing with chiral molecules and avoiding "hallucinations" of order in disordered regions. To address these issues, the model incorporates confidence measures to assess the accuracy of its predictions. Overall, the introduction of AlphaFold 3 is set to change the landscape of biological research by providing a more comprehensive and accelerated method to understand the molecular machinery of life and its implications for health and disease.

Related: DeepMind CEO Targets $100 Billion-Plus AI Drug Discovery Business With AlphaFold

AI-Powered Productivity: Transform Your Day

Revolutionize your workday with the power of ChatGPT! Dive into HubSpot’s much-anticipated guide to discover how AI can elevate your productivity and creativity.

Learn to automate tasks, enhance decision-making, and foster innovation, all through the capabilities of ChatGPT. Download the offer for actionable insights and tips on harnessing AI for a more efficient, impactful professional life.

• Efficiency Unlocked: Enhance productivity with daily ChatGPT integrations

• Everyday Solutions: ChatGPT for streamlined email and project management

• Practical Insights: Guidelines for effective ChatGPT application in work

• Ethical Guidelines: Best practices for responsible ChatGPT use

• Future Ready: Skill enhancement and learning with AI tools

Stack Overflow bans users for rebelling against OpenAI partnership

Stack Overflow is facing significant backlash from its community following a partnership announcement with OpenAI to use the site's content to train AI models like ChatGPT. Many users, in an attempt to protect their contributions from being used in AI training, have edited or deleted their highest-rated answers, leading to suspensions and bans by the site’s moderators. This has sparked a debate about the ownership and use of user-generated content, especially since Stack Overflow’s terms of service claim irrevocable rights to all user contributions.

In response to the partnership, some Stack Overflow users have suggested that OpenAI should transparently cite the sources of the answers used by ChatGPT, arguing this would add credibility to the responses and clarify the AI’s decision-making processes. This controversy has highlighted the broader implications of how user data is leveraged for profit by tech companies, and the community's growing discomfort with these practices.

Additionally, the situation at Stack Overflow reflects a larger trend in the tech industry, where companies are increasingly reversing their policies on AI use due to potential profits. This shift is part of a broader, rapid embrace of generative AI technologies across various platforms, including gaming and hardware, underscoring the lucrative yet controversial nature of AI development and its expansive impact on digital content creation and management.

Related: Stack Overflow Upset Over Users Deleting Answers After OpenAI Partnership

OpenAI posts Model Spec revealing how it wants AI to behave

OpenAI is providing insight into the rules and guidelines that direct its conversational AI models, like ChatGPT. These models don’t have inherent limits, making them versatile yet prone to errors and manipulations. OpenAI's "model spec" outlines a set of high-level rules and behavior guidelines that help the AI navigate complex requests and ethical dilemmas, such as refusing to generate false claims about public figures or biased product recommendations. This approach aims to balance the AI’s responses to maintain objectivity and prevent misuse.

The model spec includes meta-level objectives and specific instructions crafted in natural language that guide the AI's responses according to the developer's intent. For example, a version of ChatGPT might solve a math problem step-by-step rather than giving a direct answer if so programmed by its developer. This level of control helps prevent the AI from engaging in topics or actions that are outside its designated scope, like discussing unrelated historical events or creating inappropriate content.

OpenAI’s publication of its model spec offers transparency on how it manages its AI’s interactions and sets boundaries. This initiative not only aids developers in understanding and setting their own AI’s limits but also helps users grasp why and how an AI might refuse certain requests. However, enforcing these guidelines is challenging and the system isn't foolproof, as users often find ways around these safeguards or encounter unforeseen scenarios that the AI isn't prepared to handle.

Other stuff

• Meet AdVon, the AI-Powered Content Monster Infecting the Media Industry

• RAFT: Sailing Llama towards better domain-specific RAG

• Improving GPT 4 Function Calling Accuracy

• OpenAI destroyed a trove of books used to train AI models. The employees who collected the data are gone.

• How prepared are we for deepfakes? Researchers call for a shift in AI to protect women

• OpenAI’s flawed plan to flag deepfakes ahead of 2024 elections

• Udio launched Audio Inpainting

• Autodesk's AI turns text or still images into 3D models

• What’s up with Llama 3? Arena data analysis

• OpenAI Is ‘Exploring’ How to Responsibly Generate AI Porn

Hope you enjoyed today's newsletter

⚡️ Join over 200,000 people using the Superpower ChatGPT extension on Chrome and Firefox.