AI Secrets is your weekly update on how AI is changing our everyday lives. Our experts will keep it clear and simple, so you can stay ahead of the game. This week we are focussing on the return of AI hype.Share this with anyone you want to keep up to date.
Record breakers & truth seekers amp up the hype 🥇🕵️
There’s been a lull in AI hype of late; things have been quiet ever since Apple’s muted developer conference just over a month ago. That all changed last week, as seemingly every big AI player launched a major new product or update. OpenAI led the charge with Code Interpreter, which gives ChatGPT the power to run Python code in order to fulfill your prompts. Not to be outdone, Google has finally made Bard accessible to EU users, and celebrated the feat with a new update that adds Google Lens support to allow Bard to interact with images, although it sadly did not perform smoothly when we tested an early version.
Elsewhere, Anthropic - the plucky AI underdogs with a multibillion-dollar valuation - had a launch party of their own for Claude 2, the new and much-improved version of their own large language model. As is now obligatory, Claude is multimodal (it’s learned how to interpret images and wrangle spreadsheets), although much like Bard we encountered some teething problems in early testing of the image recognition. Claude brings other tricks to the table though, including the largest “context window” of any commercial AI currently available. The context window is similar to human memory, and defines the amount of information the AI can process at one time. Claude can process 100,000 “tokens”, roughly equivalent to 75,000 words, or the length of the first Harry Potter novel. It’s also approximately three times larger than GPT-4’s context window and this expanded memory should help to reduce hallucinations and provide Claude with a surprising edge.
While Anthropic is breaking records, Elon Musk is chasing loftier goals. This week saw the launch of xAI, his new AI venture whose stated mission is no less than “to understand the true nature of the universe”. To help crack this case, Musk has gathered a team of 11 of machine learning’s top academics, but as many have noted this is still a relatively small posse when compared to the legions of top minds assembled by Silicon Valley’s biggest beasts. These rivals are not seeking to uncover universal truths, however, and it will be interesting to see where Musk’s latest mission takes him, although it’s unlikely that he’ll arrive there any time soon.
In the meantime, exciting things are happening with AI here on Earth, so don’t miss out on our in-depth look at the remarkable ChatGPT Code Interpreter model, and how you can use it to start doing coding tasks with zero coding knowledge!
Beta Testers Wanted!!
SPYGAMES is the thrilling new experience where you’ll jump, climb, throw and dodge in fun immersive challenges developed with CIA experts to stretch your physical and mental agility. Click the link below to get EARLY access as a beta tester.
SPYGAMES is coming to SPYSCAPE in Manhattan, Summer 2023
Scientific American takes a look at the latest AI advances in drug testing, which is leading to huge gains in the battle to develop new ways to combat increasingly antibiotic-resistant superbugs.
The Hollywood actor’s strike seemed to worsen as actors claim they’ve been offered a rights package on AI reproductions of their work that leaves them “in jeopardy of being replaced by machines”
Google’s dedicated medicine model, Med-PaLM 2, is currently being tested at hospitals including the Mayo Clinic. Results are encouraging, if you’re able to ignore the occasional hallucination.
Adobe joined in with the product launch frenzy by making its text-to-image service, Firefly, globally available, while boasting that their AI has already generated a billion images since launching in April.
Researchers at Rice University confirmed the perils of training AI on AI output; they looped outputs through the same model five times and triggered what they call “Model Autophagy Disorder”, or MAD.
A mother snapped her son posing with mannequins at a Sydney exhibition and was so pleased with the result she submitted it for a photography prize. It was rejected for looking “a little AI-ish”!
