• Future Perfect
  • Posts
  • #115. 😱 Twitter / X allows 3rd party AI training on posts 🤯 New Nvidia model crushes ChatGPT-4 🎙️ Google NotebookLM's podcasts can now be customized

#115. 😱 Twitter / X allows 3rd party AI training on posts 🤯 New Nvidia model crushes ChatGPT-4 🎙️ Google NotebookLM's podcasts can now be customized

Plus, AI tools: Learn new languages with Speak | Do anything with your voice with Flow | Glean is AI for all your work data

🎙️ Check out my new Marc Hoag AI Law Pod (formerly my 200-episode podcast Autonomous Cars with Marc Hoag). For this inaugural relaunch — the 201st episode — it’s “hosted” by Google's NotebookLM for added context and engaging talks about each issue of Future Perfect. Let me know what you think and whether I should keep doing it! (Available on all major podcast platforms; Spotify is first, Apple Podcasts is often delayed.)

❓ So, did you like the podcast? Should I keep doing it?

Login or Subscribe to participate in polls.

  • X Updates Privacy Policy for AI Use: Social platform X (formerly Twitter) has updated its privacy policy to allow third-party collaborators to train AI on user data unless users opt out. This follows owner Elon Musk's use of X data to train xAI’s chatbot, Grok. (Review here)

  • Licensing Data for Revenue: X, like Reddit and other media organizations, may be exploring licensing data to AI companies as a new revenue stream.

  • Opt-Out Unclear Until November 15: While the policy mentions users can opt out of sharing their data, the specific option will not appear until November 15 when the policy becomes effective.

  • Data Retention and Removal Changes: X will now retain data for varying periods, depending on legal and operational needs, and warns that public content may persist even after removal.

  • Scraping Penalties Introduced: A new “Liquidated Damages” section in X's Terms of Service imposes penalties of $15,000 per million posts scraped within a 24-hour period, as part of efforts to curb unauthorized data scraping.

  • Nvidia’s New AI Model Surpasses GPT-4: Nvidia has quietly released an AI model called Llama-3.1-Nemotron-70B-Instruct, outperforming top models like OpenAI's GPT-4 in key benchmarks, signaling a major shift in AI leadership.

  • Strategic Expansion into AI Software: Traditionally known for GPUs, Nvidia is now emerging as a leading AI software developer, expanding beyond hardware into sophisticated language model development, challenging software-focused companies.

  • Refined Open-Source Model: Nvidia refined Meta’s Llama 3.1 model using advanced techniques like Reinforcement Learning from Human Feedback (RLHF), producing a highly capable, cost-efficient model for businesses.

  • Business Flexibility and Risks: While offering significant potential for enterprise use, Nvidia cautions that the model isn’t optimized for specialized domains like math or legal reasoning, urging businesses to implement safeguards.

  • AI Competition Intensifies: Nvidia’s push into AI software has escalated competition, challenging proprietary systems and pushing other tech giants to accelerate their research and development.

🎙️ New in NotebookLM: Customizing your Audio Overviews and introducing NotebookLM Business

  • New Audio Overview Customization in NotebookLM: Users can now customize NotebookLM's Audio Overviews, tailoring the focus and expertise level of the AI hosts to suit specific needs. This allows for more personalized deep dives into uploaded materials.

  • Background Listening Feature: NotebookLM now supports background listening, enabling users to explore their sources and gather citations without pausing the audio overview.

  • NotebookLM Business Pilot Program: Google introduced NotebookLM Business, a version tailored for businesses, universities, and organizations, with enhanced features focusing on data privacy and security.

  • Early Access to NotebookLM Business: Organizations can apply for the NotebookLM Business pilot to access new features, training, and email support, with pricing and general availability expected later this year.

  • NotebookLM Gains Popularity: Already used by over 80,000 organizations, NotebookLM continues to grow, offering a versatile tool for understanding and engaging with complex information.

✨ PowerPrompt™

I want you to roast me based on our conversations and what you know about me. No sugar coating. I need the criticisms to improve and upgrade myself.

As a complement — but probably not a compliment — to yesterday’s PowerPrompt™, this is a logical followup, but definitely probably brace yourself for a rude awakening. Again, this will only work with ChatGPT.

🧰 AI Tools & Resources

  • Speak: Learn new languages with AI.

  • Flow Voice: Use your voice to do anything.

  • Glean: AI to analyze work data.

🎉 THAT’S ALL FOR TODAY!

See you next time! 👋

What'd you think of today's email?

This helps me improve it for you!

Login or Subscribe to participate in polls.

Looking for past newsletters? You can find them all here.

Reply

or to participate.