What’s New in AI: Big Investments and Brainy Models

Salesforce Pours $500 Million into AI Startups

Salesforce Ventures, the money-handling branch of Salesforce, just dropped a cool $500 million into companies working on generative AI. This ups their total spend to a whopping $1 billion in just a year and a half. They’ve put money into some of the top names in AI, like Anthropic and Hugging Face.

This big spending shows just how much faith Salesforce has in AI to make waves in business software. AI is not just a fancy tool anymore; it’s reshaping entire industries. With big companies like Salesforce betting on it, we can only expect more smart developments in AI.

OpenAI Rolls Out Super Smart AI Models

OpenAI just unveiled its latest AI brainchildren, the OpenAI o1 models. These new models are like ultra-smart students that take their time to think over problems, which makes them great at coming up with thoughtful answers. They mimic how a really smart person, like someone with a PhD, would think things through step by step.

Nikunj Handa from OpenAI’s team says these models are like having a chat with a super smart PhD grad. They’re good at stuff like law, coding, and science, and they don’t need much help from us to get things right. Right now, these new models are just for developers who spend a lot at OpenAI, so not everyone can use them yet.

Legal Drama Over AI Data Scraping

The race to scoop up data for training AI is causing some legal headaches. For instance, Mumsnet, a UK parenting site, is upset that AI companies are using its content without asking. They’ve got loads of valuable data, over six billion words, that AI companies want to teach their models. Although Mumsnet tried to make official deals for their data, talks with OpenAI fell through, leading to legal issues.

This kind of drama is becoming common as AI needs more and more data to learn and get better. Sites like Mumsnet want to protect their content and get paid for it, which is fair.

Library of Congress: An AI Goldmine

The Library of Congress in the US is packed with over 180 million items, making it a goldmine for AI companies that need data to train smarter models. From old maps to recordings, this library has it all. AI companies, including giants like Amazon and Microsoft, want to tap into this resource.

Judith Conklin, who helps run the library, mentioned that they’re cool with sharing data through official channels, but scraping directly from their site messes with their services. Natalie Smith, who oversees digital stuff at the library, pointed out another issue: AI often gets things wrong when dealing with historical data, like thinking someone in an old photo is holding a smartphone.

So, here’s the scoop on how OpenAI’s latest AI models are like talking to a genius, how Salesforce is investing big in AI startups, and why historical and legal issues in data scraping matter to everyone.

FAQ: New Developments in AI and Legal Issues

1. What are the new OpenAI o1 models capable of?
The new OpenAI o1 models are advanced AI systems designed to think deeply about problems before giving solutions. They act like very smart PhD students, able to handle complex questions in law, coding, and science without much human help. These models take a bit more time to respond, ensuring their answers are well-thought-out.

2. Why is Salesforce investing so much in AI startups?
Salesforce believes that AI is the future of enterprise software, which is why it has invested $500 million more into generative AI companies, bringing its total investment to $1 billion over 18 months. This investment supports companies that are leading innovations in AI, indicating Salesforce’s commitment to integrating AI into business solutions.

3. What’s the controversy surrounding data scraping in AI development?
Data scraping involves AI companies using large amounts of publicly available data without permission to train their models. This practice has led to legal issues, like with Mumsnet, a parenting forum that accused AI companies of using its data without proper authorization. These disputes often focus on the unauthorized use of copyrighted materials to improve AI models, sparking debates about copyright infringement and compensation.

Sources Forbes