Today in AI

• Today’s big story

• News roundup

• AI research changing the world

• The tool we’re loving

• This week on the podcast

What’s happening in AI right now

The rise of reasoning AI

The AI landscape is undergoing a significant transformation as researchers and tech giants shift their focus from raw computational power to more nuanced, reasoning-based models. This evolution marks a pivotal moment in the field, with implications that stretch far beyond academic circles.

OpenAI's quantum leap in reasoning 

OpenAI has recently unveiled its "o1" family of AI models, including o1-preview and o1-mini, which represent a quantum leap in AI reasoning capabilities. These models demonstrate performance comparable to doctoral-level students in STEM fields, showcasing their ability to tackle complex problems with a level of sophistication previously unseen in AI systems.

The o1 models excel in academic and professional domains, particularly in physics, chemistry, and biology. What sets them apart is their enhanced reasoning abilities and stronger safety protocols. This development signals a shift in AI research priorities, moving from scaling up model size to improving fundamental cognitive functions.

Google's fact-checking innovation

As AI models become more advanced, ensuring their accuracy and reliability becomes increasingly critical. Google has taken a significant step in this direction with DataGemma, an AI fact-checking tool designed to ground large language model responses in verifiable data. By using techniques like Retrieval-Interleaved Generation (RIG) and Retrieval-Augmented Generation (RAG), DataGemma cross-references AI-generated content with Google's Data Commons, aiming to reduce hallucinations and improve factual accuracy.

While still in its early stages and facing limitations in data availability and accuracy, DataGemma represents a crucial step toward creating more trustworthy AI systems. Its success could set new standards for fact-checking in AI and have far-reaching implications for industries relying on AI-generated content.

The AI "subconscious"

As we develop more sophisticated AI models, researchers are drawing intriguing parallels between large language models (LLMs) and human subconscious processes. The hidden layers within LLMs mirror aspects of human subconscious decision-making, employing probabilistic shortcuts similar to human heuristics.

This comparison offers fascinating insights into both artificial and natural intelligence. It highlights how AI can inherit biases from training data, influencing responses in ways reminiscent of human intuition. Moreover, understanding these "subconscious" processes in AI is crucial for ethical AI development and deployment.

The journey of AI development is far from over. As we continue to unravel the mysteries of machine intelligence, we may find ourselves gaining new insights into human cognition as well. The future of AI is not just about creating smarter machines, but about deepening our understanding of intelligence itself.

News roundup

The top stories in AI today.

NEW LAUNCHES

The latest features & products in AI innovation.

• Zoho Releases AI-Powered Analytics to Boosts Business Intelligence

• NVIDIA and Oracle Unveil Zettascale AI Supercomputing for Enterprises

• Apple Has Shared the First Image from Its Upcoming Image Playground Feature

• Hume’s AI Assistants Bring Emotional Intelligence to LLMs

• New Gemini Feature from Google Lets You Create Your Own AI Chatbot

• Apple Unveils Private Cloud Compute for Enhanced AI Security

• Bumble to Harness AI to Boost Dating Success and User Experience

GADGETS

Computers, phones, wearables & other AI gizmos.

• Smartphone Makers are Increasing RAM to Accommodate On-Device AI

• What to Know about Apple AI Features and When You Can Use Them

• Echelon, Skyfire Merge to Launch AI-Powered Drones for Public Safety and Defense

• Meta Rolls Out AI Features for Quest Headsets

MODELS

Deployment, research, training, infrastructure

• Google’s AI Fact-Checker Aims to Curb Hallucinations

• Mistral AI Unveils Pixtral 12B, Its First Multimodal AI Model

• Project Strawberry Is Here: OpenAI Drops ‘o1’ AI Model That Reflects Before Acting

• OpenAI’s New ‘o1’ AI Models Are As Capable as PhD STEM Students

• Do AI Models Have a Subconscious?

• New Research Breakthrough Makes Neural Networks More Understandable

• Jina AI Unveils Compact Models for Superior Web Content Processing

• A New Chinese Open-Source AI Is Giving Siri and Alexa a Run for Their Money

GOVERNMENT

Press releases, regulation, defense & politics.

• Tech Leaders and White House Convene to Tackle AI’s Energy Challenges

• EU Watchdog Probes Google’s AI Model for Privacy Risks

• Yann LeCun and Geoffrey Hinton Clash Over AI Safety Bill SB 1047

• NYC to Introduce Legislation Requiring List of Approved AI Tools

ENTERTAINMENT

Content creation, movies, tv, music, gaming & sports

• Sony to Invest Billions into Original Content with Keen Eye on Anime and AI

• NVIDIA’s Holoscan Allows Developers to Integrate AI Into Live Media

• AI-Generated Tribute to Czech Composer Dvořák’s Showcases in Washington

• Google’s NotebookLM AI Turns Notes into Engaging Podcasts

• Facebook and Instagram Make AI labels Less Prominent on Edited Content

• Adobe is Positioning Firefly Video as Essential Tool for Modern Editors

AI research changing the world

The latest breakthroughs and most pivotal papers — broken down in language anyone can understand.

Getting 26,500 LA City Workers to Get Al Training

The Information Technology Agency of LA has unveiled an ambitious plan to integrate Al across all city departments, potentially revolutionizing public service delivery for 4 million residents. At its core, the roadmap outlines a strategy to train over 26,500 city employees in Al, starting with a 3-month trial of Google Gemini Enterprise. But it's not just about efficiency-it's a blueprint for the future of urban governance. The plan addresses critical ethical concerns head-on, prioritizing responsible Al adoption to mitigate risks of bias, privacy breaches, and job displacement. By 2028, cities fully embracing this Al-driven approach could see a 15% reduction in traffic congestion and a 10% improvement in air quality. As Al becomes integral to city operations, it raises intriguing questions: How will this reshape the relationship between citizens and local government? Will it lead to more trust or skepticism? And as other cities inevitably follow suit, could we be witnessing the birth of a new era of hyper-efficient, Al-augmented urban centers?

The AI tool we’re loving right now

The best way to get AI literate? Try the tools!

BigVu

There’s no way around it — being a content creator, influencer or business leader these days requires publishing high-quality video content. That’s why BigVu is such a powerful tool. BigVu uses the power of AI to allow you traverse the entire lifecycle of video creation and publishing with some very cool features:

• Eye contact correction

• Subtitle & caption generation

• Video-to-text transcription

• Script generation

• Teleprompter

• AI voice changing & cloning

• Background music generator

• Post Scheduler

• Landing page generator

• And much more

The best part? CO/AI subscribers get access to a special discount when they sign up using the link below. Use our promo code MEHFIN63 at checkout to receive $20 off when you subscribe.

Want to learn how to use BigVu? Check out our tutorial here.

This week on the podcast

Can’t get enough of our newsletter? Check out our podcast Future-Proof.

Get more with a Pro account

Paid members get access to discounts on AI tools, expert-written tutorials and deep industry data and leaderboards.