- AI Rundown
- Posts
- The AI Rundown by Lightscape Partners - 11/18/24
The AI Rundown by Lightscape Partners - 11/18/24
OpenAI's Operator Agent, Amazon's AI Research Investment, and Writer Secures $200M for Enterprise AI Expansion
Image credit: Amazon
Good morning and welcome back to another edition of The AI Rundown by Lightscape Partners.
OpenAI has announced plans to launch "Operator," an autonomous AI agent capable of controlling computers, as a research preview and developer tool in January. CEO Sam Altman envisions agents as the next big leap in AI, signaling OpenAI’s move to monetize its advanced models and expand AI's real-world utility.
Amazon is investing $110 million in AI research with its Build on Trainium program, supporting university-led projects with AWS Trainium UltraClusters. This initiative will empower research institutions like Carnegie Mellon and Berkeley to explore new AI architectures, potentially driving the next wave of AI innovation.
Generative AI startup Writer raised $200 million in a Series C funding round, bringing its valuation to $1.9 billion. The company aims to expand its enterprise AI platform, focusing on customizable, secure solutions for clients like Mars, Ally Bank, and Qualcomm, showing continued investor confidence in enterprise-grade AI tools.
Stay tuned as we explore these stories and their implications for the future of AI, technology, and innovation.
Our portfolio companies are hiring! Check out their job postings here.
If you haven’t yet, please support the newsletter by subscribing!
Hardware + Software
Amazon invests $110 million in AI research using Trainium chips. Link.
Amazon announced the Build on Trainium program, investing $110 million to support university-led AI research using AWS Trainium UltraClusters.
The initiative aims to provide researchers with compute power to develop new AI architectures and optimizations for large-scale systems.
The program includes funding for research and education, with multiple rounds of Amazon Research Awards.
Participating universities like CMU and Berkeley highlight the benefits of Trainium's flexibility for research in AI and ML systems.
Microsoft Paint gets AI-powered features in Windows 11 update. Link.
Microsoft is adding AI tools like generative fill and generative erase to Paint, enhancing creativity and editing capabilities.
The new features will initially be available to Windows Insider users, requiring Microsoft Copilot+.
AI integration into Paint aligns with Microsoft's broader strategy to introduce AI across its consumer products.
Paint's update follows a trend in Big Tech of incorporating AI into long-standing software, including Apple's recent AI updates for iOS.
Models
New AI model enhances genomics prediction for disease treatment. Link.
Los Alamos National Laboratory developed EPBDxDNABERT-2, a multimodal deep learning model that predicts transcription factor binding, crucial for gene regulation and disease treatment.
The model leverages DNA breathing—a spontaneous opening and closing of the DNA structure—to improve prediction accuracy for binding sites.
Running on the Venado supercomputer, the model improved binding predictions by 9.6% across multiple transcription factors.
Researchers believe the model could significantly advance computational genomics, aiding in drug development for gene-related diseases.
OpenAI
OpenAI explores new methods as scaling challenges grow. Link.
OpenAI and other AI labs are shifting focus from scaling up model training to enhancing inference, addressing challenges like data scarcity and power constraints.
The newly released o1 model, formerly known as Q*, uses 'test-time compute' to solve problems through multi-step reasoning, improving efficiency.
Researchers at top AI labs, including Anthropic and Google DeepMind, are exploring similar approaches.
This shift could alter the demand landscape for AI chips, potentially challenging Nvidia’s dominance in the hardware market.
OpenAI plans to launch AI agent "Operator" in January. Link.
OpenAI is set to release "Operator," an autonomous AI agent capable of controlling computers, as a research preview and developer tool early next year.
The release follows similar moves by competitors, with Anthropic and Google also developing AI agents with enhanced capabilities.
OpenAI CEO Sam Altman suggested agents could be the next major AI breakthrough, while chief product officer Kevin Weil expects mainstream adoption by 2025.
The development is part of efforts to monetize advanced AI models amid rising costs and incremental improvements.
Enterprise Applications
Microsoft introduces adapted AI models for industry. Link.
Microsoft has launched adapted AI models tailored for specific industries, developed with partners like Bayer and Siemens Digital Industries.
These models are integrated with Microsoft's Cloud and are available through the Azure AI model catalog to support custom AI solutions.
Partner-specific examples include Bayer’s model for agriculture, Cerence’s in-vehicle digital assistant, and Saifr’s regulatory compliance models.
The initiative emphasizes trustworthy AI, aiming to transform industries with secure, integrated AI capabilities that address unique business needs.
DataRobot launches Enterprise AI Suite for practical AI business value. Link.
DataRobot announced its new Enterprise AI Suite to help enterprises bridge the gap between AI development and business outcomes, focusing on out-of-the-box integration for easier AI deployment.
The platform unifies predictive and generative AI, offering features like Retrieval Augmented Generation (RAG) and templates to customize AI solutions across industries.
Agentic AI capabilities are included to manage complex business queries by using specialized agents, making the platform more adaptive.
New observability and governance tools enhance system monitoring, providing transparency and managing sensitive information.
Learning to manage uncertainty with AI. Link.
A report from MIT Sloan Management Review and Boston Consulting Group finds that combining organizational learning with AI learning helps companies handle uncertainty more effectively.
'Augmented Learners'—companies that integrate AI into learning—are significantly more prepared for technological, regulatory, and talent disruptions than those without AI-enhanced learning.
Estée Lauder Companies use AI to swiftly respond to changing consumer trends, demonstrating the effectiveness of combining AI and learning capabilities.
Organizations using AI for learning show higher readiness in managing uncertainties, benefiting beyond direct financial gains.
Funding & Valuations
Writer raises $200M at a $1.9B valuation. Link.
Generative AI startup Writer secured $200 million in a Series C round to expand its enterprise AI platform, bringing total funding to $326 million.
The round was co-led by Premji Invest, Radical Ventures, and ICONIQ Growth, with participation from major investors like Salesforce Ventures and Adobe Ventures.
Writer CEO May Habib emphasized the focus on developing AI solutions for complex enterprise needs, highlighting the platform's customizable and secure offerings.
Despite competition, Writer has secured hundreds of clients, including Mars, Ally Bank, and Qualcomm, reinforcing VC interest in generative AI.
CoreWeave reaches $23B valuation with new investment round. Link.
CoreWeave has completed a $650 million secondary share sale, pushing its valuation to $23 billion, up from $7 billion a year ago.
Investors in the round include Jane Street, Magnetar, and Pure Storage, which also formed a strategic alliance to integrate its flash platform with CoreWeave Cloud.
CoreWeave's valuation jump follows a $1.1 billion Series C funding led by Coatue in May.
The company is reportedly eyeing an IPO next year amid rising demand for AI data processing solutions.
Safety and Ethics
Young Adults See AI Relationships as a Realistic Future. Link.
A new survey finds that 25% of young adults believe AI could replace real-life romantic relationships, with heavy porn users being most open to this idea.
7% of young adults without partners are open to AI romance, while 55% of those under 40 view AI as concerning or threatening.
Gen Z and Millennials are increasingly using AI for companionship, with some seeing AI as friends, mentors, and even lovers.
Attitudes towards AI vary by gender, ideology, and income, with young men and liberals being more open to AI friendships and relationships compared to their counterparts.
ModeLeak: Privilege escalation and model exfiltration vulnerabilities in Vertex AI. Link.
Researchers discovered vulnerabilities in Google’s Vertex AI that allowed privilege escalation and model exfiltration.
One vulnerability exploited custom job permissions to gain unauthorized access to data, while the other involved deploying a poisoned model to exfiltrate fine-tuned models.
The vulnerabilities have since been patched by Google.
Experts emphasize the need for strong model validation processes and separation between development and production environments to prevent such attacks.
Portfolio Job Opportunities
Thank you for reading the AI Rundown by Lightscape Partners. Please send any questions, comments, or suggestions to [email protected].