Hello, AI Enthusiasts! (And the rest of y’all!) Welcome to the latest edition of AI Insider, where we bring you the most recent and exciting developments in the world of artificial intelligence, with a special focus on generative AI. Our goal is to make this fascinating technology accessible and enjoyable for everyone, so let’s dive in! (Written and compiled by human Michelle Johnson and chatbots ChatGPT, Bing and Bard.)
AI In the News
By Michelle Johnson
Google announced a boatload of AI-themed updates at Google I/O, their annual developer conference that happened on the left coast this week. We’ll get to that in a minute.
First, a tiny bit of context. Google’s been under fire from critics alleging that they are falling behind on the AI front. Their chatbot, Bard, launched to yawns in the wake of ChatGPT and Bing. To be fair, Bard started out as a bare-bones beta. My opinion: You can’t count Google out yet.
Based on what they showed in the opening keynote, the fact that they have been immersed in AI for YEARS, and that some of us use their products every flipping day (market share!), read on to see why.
Here’s what’s launched and coming soon:
(Disclaimer: Most of this stuff hasn’t dropped yet. Don’t flame me if you don’t see it. I’m just the messenger.)
- Gmail will include a feature called “Help Me Write.” It works similarly to GhatGPT or Bing when you ask it to write an email. Of course, if you use Gmail, it will be insanely convenient to just do this right in your inbox.
- Bard no longer has a waitlist! Dive right in. The biggest surprise: it will generate images using Adobe Firefly! Whaaat? Adobe and Google collaborating? Well, damn. (Bing uses OpenAI’s Dall-E to do the same thing, thanks to their collaboration with Microsoft.) You will also be able to upload photos to Bard. The demo showed someone uploading a photo and asking Bard to write funny captions. Bard will also export to Google Drive apps like Sheets and Docs as well as Gmail. And for those who hated Bard’s look ‘n feel, you can now switch to dark mode. There’s other stuff, like an upgrade in its ability to code, but that’s not for us mere mortals, so I won’t bother to report on that.
- Google Photos has a new feature called Magic Editor. It’s kind of what it sounds like. You can edit by magically moving elements around, and it will automatically edit in missing parts and such.
- Google Maps: Immersive mode will let you preview a route. Imagine flying a drone above the street to see where you’re going with an overlay of weather updates!
- Google Docs, the Docs “word processor” (what an old-fashioned term!), spreadsheets, and slides will be AI-ified, too. You can ask them to write text, have sheets (or Bard) generate formulas, and get this: Slides will auto-generate speaker notes! And it looks like it will include Adobe Firefly for image generation, too. As someone who hasn’t used Powerpoint for years, I love this. Too bad I’m retired, tho. I could have used this for my class lectures.
- Search: You didn’t think they would forget about search, did you? Not surprisingly, search will become more conversational, like talking to a chatbot. If you want to see that experience now, try signing up for the Bing beta.
Watch the I/O Keynote Presentations
Short Version: Google I/O in Less than 10 Minutes
Full Version. Yeah, the whole two hours!:
Learn more on this Google Labs page.
Yet More News
By Bing
Here is some of the latest AI news from Meta, OpenAI, and Anthropic that you should know.
Meta releases ImageBind
Meta, formerly known as Facebook, has been busy developing new AI models that can enhance its platforms and services.
This month, Meta announced another AI breakthrough with ImageBind2, a multisensory AI model that combines six types of data: text, audio, visual, movement, thermal, and depth. [The big whoop is that the currently hot AI image generators combine just text and images. –MJ]
One example: ImageBind could generate a realistic sound effect for a given image or video, or create a 3D model from a text description.
OpenAI debuts Shap-E
OpenAI, the research organization behind the powerful language model ChatGPT, has also been working on new AI tools that can create stunning 3D models from text or images.
Anthropic introduces Constitutional AI
Anthropic is a new AI startup founded by former OpenAI employees who want to create more aligned and trustworthy AI systems. In March, Anthropic introduced Claude5, a chatbot that uses a technique called Constitutional AI to guide its behavior. [This week the company released a copy of this “constitution.” Now, that’s transparency. — MJ]
[Update! This morning I see that Claude’s ability to read and analyze documents has been significantly upgraded. It can gobble up an entire book in less than a minute. It can analyze business docs in seconds. Details here.]
Constitutional AI involves conditioning an AI model with a set of principles that reflect its values and goals. Anthropic’s constitution draws on sources such as the United Nations Declaration of Human Rights, Apple’s terms of service and Anthropic’s own research principles.
Anthropic claims that Constitutional AI can help prevent AI models from producing harmful or biased outputs by making them more transparent and accountable.
****************************************
Fun Stuff to Try
Take this free course: Learn Prompting: Your Guide to Communicating with AI
There’s a new AI chatbot on the block (See above). It purports to be kinder and gentler. Meet Claude. https://www.anthropic.com/index/claudes-constitution
Check out PDF.ai. You upload a pdf and then ask it questions about the document.
****************************************
Thank you for reading AI Insider, the newsletter that keeps you informed and inspired by artificial intelligence. We hope you enjoyed this edition and learned something new. Until next time, stay curious and keep exploring the amazing possibilities of AI.
Excellent work Michelle!