The Generative AI News (GAIN) brief on April 27, 2023 was another week of heavy news. We’ve reviewed the top stories about GAIN and what they mean for the industry. Developments include news from ChatGPT, HuggingFace, Google, Nvidia, Sensory, Hour One, D-ID, Deepfake his musicians and more.
today’s host Brett Kinsella and Voicebot.ai Eric SchwartzHere are this week’s top stories in Generative AI Land:
- new vision plugin: OpenAI’s Greg Brockman demoed some of the new ChatGPT plugin features. Some are amazing.
- The Promised “Super App” Virtual Assistant: Brockman’s demo and discussion of product philosophy provide insight into where ChatGPT is headed. Go, Alexa. Go away, Siri. ChatGPT may be the virtual assistant we’ve always wanted.
- ChatGPT is not secret. While ChatGPT seems to be getting a lot of public attention, OpenAI understands that not all users want all chat conversations to be permanently stored and used for future model training. Recognized. Incognito mode (i.e. private chat) is now available, and “business mode” will be available soon.
- ChatGPT Open Source Contest: Hugging Face stepped up and offered an alternative to ChatGPT built on open source models and data. It’s a smaller AI model than ChatGPT, and much better.
- Drake, The Weeknd, Bad Bunny, Rihanna Are Talking: Viral hits from big stars are common. Deepfake viral hits that mimic the voice, style, and appearance of big stars can also become commonplace. Ghost Rider 777 attacks again!
- Joe Rogan’s comments are deep: In a new deepfake that mimics Joe Rogan’s podcast, comedians and commentators talk about “slippery” slopes.
- Grimes jumps on board: The music artist says she will split royalties 50/50 with anyone who deepfakes her voice. She has no label and no binding legal restrictions, which makes her more flexible than most musicians.
- prompt to video: Hour One has introduced a new text-to-video solution that enables complete video generation for presentations from a single prompt.
- Canva wins digital people: D-ID has introduced a new Canva app that lets you add generative video to any project.
- Chatbots skyrocket in popularity: Character AI has raised $150 million in funding at an obscene valuation. With virtualized Elon Musk, Mark Zuckerberg, and his 2.7 million unique chat avatars, in just two months he has reached 100 million users.
- Bard learns to code: Google is slowly catching up to the leader in generative AI. ChatGPT competitors, or Bing Chat competitors can now code. This is not yet a true competitor to GitHub Copilot.
- Sec-PaLM enters security: Google also rolled out a new cybersecurity solution with the modest name of Google Cloud Security AI Workbench. It is based on a fine-tuned version of the PaLM Large Language Model (LLM).
- ChatGPT dominates: Sensory has rolled out a new hybrid on-device and cloud solution that allows you to use ChatGPT and similar services on your device.
- NVIDIA on rails: Nvidia’s LLM, NeMo, has added new features to add guardrails to other LLMs to align model output to corporate safety and security requirements. NeMo Guardrails is open source and designed to work with any LLM.
The show will be recorded live and streamed Thursday at noon ET via YouTube and LinkedIn. You can revisit our weekly discussions on Voicebot’s YouTube channel. Join us live on YouTube or LinkedIn next week. Also, comment and join the live show. We will likely give you a shout out and may even put your comments on screen.
Generative AI News Brief – Show Notes April 27, 2023
Follow @bretkinsella Follow @voicebotai
Nico Perony, Director of AI Research at Unity on Generative AI, Complex Systems and More – Voicebot Podcast Ep 314
Generative AI News – StableLM, Elon Musk, Drake Deepfake & More – Voicebot Podcast Ep 316
