- Strange Magic
- Posts
- My AI has a great personality
My AI has a great personality
PLUS: Elon Musk has a brain implant for you
Hello, Human Friend!
It was another big week for the big AI companies with OpenAI dropping a slew of updates and announcements. That isn’t stopping George R.R. Martin and other blockbuster authors from suing them for copyright infringement. At least we have spotless giraffes to cool our brains.
In this edition of Strange Magic…
👩🦱 Our chatbots are getting personal
🐉 Another round of authors sue OpenAI
📺️ Embed your name in an AI generated video
🦒 How it is even cuter without spots?
Read time: 5 minutes
🔮 What’s New
The next phase of AI hype and rapid advancement will likely be around making personal AI chatbots more personal. Already, the second most popular chatbot platform is character.ai, a vast world of bots, each with a unique personality. Rumors are swirling that Meta will soon launch a similar host of bots with personas. It’s no longer good enough to type a simple question into a box and get an often eloquent and seemingly rational typed answer from a computer.
OpenAI’s response to the personality rush came last week with major updates to the ChatGPT interface. First, ChatGPT can now listen to your voice and respond with its own voice. Perhaps people will spend more time with ChatGPT if they can just talk to it. But ChatGPT can also see now… sort of. The GPT 4 model can now process what’s in an uploaded image and talk about it with you. This means, it might be able to tell you if you have poison ivy in your yard, just by uploading a picture of it. But, please don’t upload any pictures of your rash, at least not for now.
Taking the seeing metaphor one step further, OpenAI also announced the upcoming release of DALL-E 3, the next version of its popular text-to-image generator. With DALL-E 3, OpenAI has rethought the image generation process, building the software on top of ChatGPT. This allows users to describe exactly what they want to see, allowing ChatGPT to write an appropriate prompt for it. Pour one out for all the prompt engineers out there.
OpenAI must also be looking over its shoulder at Google who is planning to release a brand new multi-modal (“seeing and hearing”) modal called Gemini in the fall. Reviewers suggest it could rival OpenAI’s leading GPT 4 model.
In Other News
If you can’t beat them: In 2022, Getty Images took a stand against AI generated images, banning them from the stock platform, citing copyright concerns. In Feb. of this year, they went a step further, suing Stability AI claiming copyright infringement against the work on their site. Now, they have unveiled a proprietary image generator of their own, trained exclusively on approved Getty content. Like Adobe, Getty is claiming all generated images will be copyright safe. They are even including indemnity in the event of any legal issues arising.
You know nothing, ChatGPT: George R.R. Martin is just one in a group of high profile authors, including John Grisham and Jodi Picoult, to sue ChatGPT’s parent company OpenAI for copyright infringement, claiming the multibillion dollar startup trained powerful AI models like ChatGPT on their work without permission or licensing agreements. And, that these models threaten their livelihoods. This adds to a growing stack of similar suits against most of the major players in the AI space. OpenAI maintains that they are working hard to understand creators’ concerns over AI. One guess, being wiped out by a supercomputing mimic.
They’re putting chips in our brains: Don’t be silly, no one is putting chips in our… oh wait. Well, it looks like Elon Musk’s “brain-computer-interface” company is looking for test subjects in its first human trials of the Neuralink system, software, and robotic implant surgeon (everyone’s next Halloween costume). While it is a big step for the company hoping to help quadriplegia patients, it’s worth noting that Neuralink joins a growing number of research firms testing and reporting promising results in the field of brain-computer-implantation.
📖 Good Reads
Sequoia Capital offers up a comprehensive look at today’s generative AI landscape, the major companies involved, and what we might expect in the near future. The hypothesis… things are going to keep moving very fast.
Chatbots have a confidence problem. That is, they are often too confident when giving incorrect answers. No one wants to be that guy, so Google is working to make their AI’s less confident, and maybe a little more accurate in return.
More Reads
⚙️ Try These
Pika Art Encrypt: Pika is primarily a text-to-video generator, but their newest feature is already going viral. They call it “Encrypt”, and it allows users to input text or upload an image to be embedded (sometimes almost subliminally) into a generated video. Join the Pika Labs beta to try it out.
Pinokio: If you are looking to try the latest open source generative AI tools but don’t have the know-how to install confusing packages, Pinokio might be for you. The “AI Browser” installs AI models with a single click. (This is an unverified tool, so please do your research before installing.)
✨ We’re Diggin’
It’s out of our hands: When Base Design was developing looks for the newest program at Brussels opera house La Monnaie, they decided to take a cue from the theme of Fate, by leaving the looks up to fate and a handful of generative AI tools. The team used story summaries and a curative knack to pull the full package together.
Knowing is half the battle: Google DeepMind continues to stand at the frontier of AI in scientific discovery. Previously scientists had classified about 0.1% of genetic mutations that can lead to disease. Google claims their new system has pushed that classification to about 89%, potentially allowing for much quicker testing and drug development.
🌤️ Fresh Air
Another spotless giraffe spotted: After an extremely rare spotless giraffe was born at a Tennessee zoo last month (believed to be the only one in the world), another spotless giraffe has been spotted in the wild in Namibia. It’s just so hard to stand out these days.
Let us know how we’re doing, and drop of a note so we can make Strange Magic better for you.
Well, that’s enough strangeness for one day. Hope you learned something new. If you enjoyed it, please spread the word! Tell your friends to subscribe!
And, of course, if you have anything interesting to share, or just want to say hi, please feel free to reach out to us.