AI TRENDS
News & Breakthroughs
Apple plans to unveil new iPhones, AirPods, and Apple Watches on September 10 (Mark Gurman / Bloomberg)
Meta's "Lama 3.1 Impact Grants" program seeks proposals from nonprofits that leverage Meta's AI model to create positive societal impacts. The grants are intended to support projects or services that utilize this AI technology for the greater good.
Case Studies & AI Research
MuMA-ToM: Multi-modal Multi-Agent Theory of Mind (See this Charity Spring commentary today explaining multimodal AI)
Charity Spring blog post: 4 case studies of charities utilizing OpenAI.
AI TOOLS
Marketing
Content Creation
YouTube Music can now find songs when you hum or whistle into the app.
Video
Vibeo: AI-enhanced platform for collecting and using video testimonials.
Google Deepmind just introduced Veo, their most capable model for generating high-definition video. and Imagen 3, our highest quality text-to-image model. We’re also sharing new demo recordings created with our Music AI Sandbox.
DreamFactory: Pioneering Multi-Scene Long Video Generation with a Multi-Agent Framework
This research paper introduces "DreamFactory," a new method for creating longer, multi-scene videos that are consistent in style and content. While current models are good at making short video clips, they struggle with longer videos. DreamFactory uses a combination of techniques to overcome these challenges, ensuring the videos maintain a consistent look and feel. The researchers also developed new ways to evaluate these longer videos and provided a dataset of over 150 videos for further research.
Images
Midjourney's AI-image generator website is now officially open to everyone - for free (Lance Whitney / ZDNET)
Black Forest Labs’ new AI image generator called Flux might actually be better than Midjourney & Stable Diffusion. There are three versions of Flux.1: Pro, Dev, and Schnell, in decreasing order of size.
OpenAI now lets ChatGPT’s free users create 2 images per day with DALL-E 3.
Microsoft added AI image generation to Paint.
Google Deepmind just introduced Imagen 3, their highest quality text-to-image model. They are also sharing new demo recordings created with their Music AI Sandbox.
Text
Fundraising
Bots to Agents: The Next Great Leap Forward Is Autonomous AI. This new Salesforce Einstein Agentforce Platform is a groundbreaking development in AI, transitioning from basic chatbots to fully autonomous AI agents. These advanced agents, such as the Einstein SDR Agent and Einstein Sales Coach Agent, can autonomously manage tasks like engaging with potential donors, addressing questions or concerns, and simulating donor conversations to better prepare fundraising teams. By automating routine interactions, these AI agents allow nonprofit staff to focus on building deeper relationships with supporters, ultimately enhancing the organization's ability to raise funds and advance its mission. This innovation represents a significant leap forward in how AI can support and scale nonprofit fundraising efforts, enabling more personalized and effective donor engagement.
Donor Analytics
Copilots and Operational Efficiency
Zoom launched AI features for meeting summaries, video backgrounds, event landing pages, & the ability to convert text in images into editable formats.
Data Management & Visualization
Notion’s ‘Charts’ enables you to visualize your data with one click.
Engagement
Chatbots & Virtual Assistants
Microsoft added a Copilot bot to Telegram & AI image generation to Paint.
OpenAI for Nonprofits recently introduced discounted access to ChatGPT Team for $20 per month per user, with larger nonprofits eligible for a 50% discount on ChatGPT Enterprise, providing advanced AI models, tools, team management features, and strong privacy and security standards.
Social Media
Meta just announced new Threads web features including insights, multiple drafts and scheduling functions to make it easier to plan content and connect with new and existing audiences.
COMMENTARY
Charity Spring
The next AI trend for 2024 is likely to be multimodal AI and the optimization of smaller models (see commentary below re: China)
Other Commentary
Q&A with AI pioneer Yoshua Bengio on advocating for AI regulation, his support for California's AI safety bill, AI achieving human-level intelligence (Shirin Ghaffary / Bloomberg)
Can Artificial Intelligence Embody Moral Values?
This research paper challenges the neutrality thesis, which argues that technology is inherently value neutral. The authors contend that this view, typically applied to traditional technologies, does not hold when applied to AI, especially in high-stakes areas like healthcare and policing. They argue that AI, particularly autonomous agents, can embody moral values through their computational models. The paper provides a conceptual framework and empirical evidence from text-based games, showing that AI with moral models exhibits more ethical behavior, thus contradicting the idea that all technologies are value-neutral.
Future Forecasts