- Daily Zaps
- Posts
- AI deep fakes and 'weaponized weather'
AI deep fakes and 'weaponized weather'
AI and weather, Google's answer to Dall-E, AI discovers viruses, OpenAI AI benchmark tool, and more
Welcome to Daily Zaps, your regularly-scheduled dose of AI news ā”ļø
Hereās what we got for ya today:
āļø AI deep fakes and 'weaponized weather'
š¼ļø Googleās text to image available to everyone
š§¬ AI helps discover new virus species
š¤ OpenAI releases AI benching marking tool
Letās get right into it!
CLIMATE
AI deep fakes and 'weaponized weather'
In the aftermath of Hurricane Helene, social media has become a hotspot for both AI-generated and human-driven misinformation, spreading false narratives about the recovery efforts and fueling new fears about Hurricane Milton, which recently hit Floridaās Gulf Coast. Conspiracy theories, amplified by comments from public figures, and AI-generated deepfakes, such as images of Trump and Biden in the flood zones, have circulated widely.
The American Red Cross and other agencies have launched efforts to counter this disinformation, emphasizing the importance of accurate information from trusted sources. Despite the challenges, AI is also being used positively to translate and disseminate accurate information in multiple languages, helping reach communities that may not speak English. FEMA has even published a list to address the most pervasive rumors.
BIG TECH
Googleās text to image available to everyone
Googleās AI image generator, Imagen 3, is now available for free to all Gemini users on mobile and desktop, but with a key limitationācreating images of people is restricted to Gemini Advanced subscribers. Imagen 3 is Googleās most advanced model, offering better adherence to prompts and fewer visual artifacts than its predecessors. While the free tier allows unlimited image generation at 2048 x 2048 resolution, it restricts the creation of human images.
Upgrading to Gemini Advanced, part of the Google One AI Premium Plan for $19.99/month, removes this restriction and includes additional features like 2TB of storage. Imagen 3 supports creating detailed scenes, photorealistic landscapes, and text-rendered images, though its text generation capabilities still need improvement.
CONTENT BY SALESFORGE
Fully Automated Email Outreach With AI Agent Frank
Hire Agent Frank to join your sales team and let him take care of prospecting, emailing and booking meetings for you, so your team can focus on closing deals!
Agent Frank works in two modes - fully autonomous Auto-pilot and Co-pilot, where you can review and monitor his work. And heās super easy to set up in just 4 quick steps!
He learns using first-party data you provide him during onboarding and continuously gets better as he works to book you more meetings š
HEALTHCARE
AI helps discover new virus species
Researchers have used artificial intelligence (AI) to uncover 161,979 new species of RNA viruses, vastly expanding our understanding of Earth's biodiversity. This groundbreaking study, published in Cell by an international team, is the largest virus species discovery ever recorded. Using a deep learning tool called LucaProt, the AI analyzed vast genetic data, identifying viruses previously hidden in "sequence dark matter" and revealing their diversity across various environments, including the atmosphere, hot springs, and hydrothermal vents.
The AI approach significantly accelerates virus discovery, offering a more efficient alternative to traditional methods and providing new insights into viral diversity and evolution. Researchers believe this method could identify millions more unknown viruses and be applied to studying other microorganisms, offering a deeper understanding of life on Earth.
STARTUPS
OpenAI releases AI benching marking tool
OpenAI has launched MLE-bench, a new benchmark for evaluating AI capabilities in machine learning engineering, using 75 real-world data science competitions from Kaggle. This tool assesses AI's ability to plan, troubleshoot, and innovate beyond basic computational tasks. OpenAI's advanced model, o1-preview, combined with AIDE, performed at a competitive level in 16.9% of the challenges, highlighting its progress but also its limitations in adapting to complex problems.
OpenAIās decision to make MLE-bench open-source could help standardize AI evaluations and facilitate future advancements. However, the results also emphasize that AI has not yet reached the nuanced decision-making level of human experts, pointing to opportunities for collaboration between AI systems and data scientists.
In case youāre interested ā weāve got hundreds of cool AI tools listed over at the Daily Zaps Tool Hub.
If you have any cool tools to share, feel free to submit them or get in touch with us by replying to this email.
šø Tech tidbits from around the web
How much did you enjoy this email? |