YouTube + AI...

YouTube is highly focused on AI right now

Screenshot by Sabrina Ortiz/ZDNET

⏰ Quick roundup:

  • AI.com once led to ChatGPT, now it redirects to Elon Musk's xAI, a truth-seeking AI using X's posts for training.

  • AI-driven Hollywood startups revolutionize production with efficient effects and dubbing, raising investor interest, but sparking job displacement and viewer deception concerns.

  • Developers face challenges evaluating GPT-4 and Google's AI due to evolving performance and limited training data, using other models and human judgment for assessments.

🚀 Latest developments:

Longer reads:

AI.com once took you to ChatGPT. Now, it goes to Elon Musk's xAI

AI.com used to direct to ChatGPT, an AI chatbot created by OpenAI, but now it redirects users to xAI, an AI company owned by Elon Musk. xAI claims it will work closely with X, the social media company formerly known as Twitter, and uses posts on X for training its language model. The ownership change of AI.com is mysterious, with speculations about whether OpenAI still owns it or if Elon Musk acquired it. xAI aims to be a truth-seeking machine to solve mysteries like the absence of contact with aliens. (Go deeper)

_____

The AI revolution in film and TV production: How startups are transforming Hollywood

Startups like Runway and Deep Voodoo are using AI to create special effects like de-aging and film restoration, making the process faster and more cost-effective. Other applications of AI in the entertainment industry include dubbing, recreating voices of past actors, and restoring old films and TV shows. Investors are pouring money into these startups as they see the potential for AI tech to transform how TV shows and movies are made. However, there are concerns about AI taking over human jobs and potentially deceiving viewers. (Go deeper)

____

Developers ask GPT-4 what it thinks about Google’s AI

Developers are grappling with the challenge of evaluating large-language models (LLMs) like OpenAI's GPT-4 and Google's PaLM, as these models can improve and worsen over time. Limited information from providers on their training methods adds to the complexity, leading some developers to use other LLMs for evaluation. While this approach may work well for narrow tasks with specific correct answers, it's not without its flaws, and supplementing it with human judgment remains essential to ensure reliable assessments of LLMs' performance. (Go deeper)

🧰 Tool discovery:

  • Lavender: sales email assistant that helps users to write better emails.

  • Mywave: personalizes customer interactions.

  • TLDRThis: summarizes any article instantly.

  • Robofy AI: Your personalized website chatbot.

💸 Funding news:

  • CoreWeave secures a $2.3 billion loan to expand data centers and buy more Nvidia GPUs for AI software.

  • Data-labeling startup Datasaur raises $4 million in seed round.

  • No-code data-quality startup Lightup raises $9 million in Series A funding.

  • Transcend raises $20 million in Series B funding for its generative design platform from multiple investors.

📚 More reads:

  • What can you do when AI lies about you? (NYTIMES)

  • Google begins offering free AI training courses in the UK. (ES)

  • New AI systems collide with copyright law. (BBC)

  • Bing, Bard, and ChatGPT: How AI is rewriting the internet. (The Verge)

  • When will generative AI finally make Siri and Alexa better? (Fast Company)