A week in Generative AI: Google, Stargate & EVI
A week in Generative AI: Google, Stargate & EVI
News for the week ending 7th April 2024
This week saw rumours around Google considering charging to use their search product, ChatGPT becoming accessible without an account, Microsoft and OpenAI’s plans for future cloud computing being revealed as well as an interesting new AI that can understand the nuance of human emotions.
Google considering charge for internet searches with AI, reports say
There were lots of rumours flying around this week that Google was considering changing their business model to start charging for search. That’s not quite the story here and I think it’s incredibly unlikely that Google will ever charge for search - its not their business model, would reduce the amount of data they collect and I’m sure would open up a can of regulatory worms.
Start using ChatGPT instantly
OpenAI allowing people to use ChatGPT without an account is move in the opposite direction to Google’s rumoured change. By allowing everyone free and easy access they will collect more data and provide a more viable alternative to Google’s search product in the future. I think this move is the start of some bigger changes OpenAI has planned for later this year…
Microsoft, OpenAI plan $100 billion ‘Stargate’ data centre project
There have been some big headlines around the ‘Stargate’ project this week, but its not surprising that Microsoft and OpenAI are looking at the data centres that need building over the next few years to support their AI ambitions. The most interesting thing is that the timing for when these data centres will come online broadly align with many commentators (including OpenAI’s) predictions of when we might see general artificial intelligence emerge. So ‘Stargate’ probably represents the hardware that Microsoft and OpenAI believe is required for general artificial intelligence to be developed.
Apple researchers develop AI that can ‘see’ and understand screen context
Apple really is ramping up its public contributions to AI research in the run up to WWDC this summer. I think we’re going to see a lot of interesting new AI capabilities/features from Apple that we haven’t seen anywhere else before. Seeing these features built in to one of the world’s most widely distributed operating systems will bring a lot of AI capabilities to a lot of people almost overnight, so will be a big game changer!
Navigating the Challenges and Opportunities of Synthetic Voices
It was surprising to see this announcement from OpenAI. Firstly, because they’ve had this (unannounced) capability for 2 years now, so why announce it now? The other surprising element is how good it is - being able to create an almost perfect replica of a voice from just a 15 second sample. It’s worth listening to the examples on the website, to my ear they’re indistinguishable from the original recordings!
Meet the Empathic Voice Interface (EVI) – the first AI with emotional intelligence
I’ve seen a lot of machine learning models over the past 10 years that are able to distinguish different human emotions from either video, audio or both so not a surprise to now see this capability surfacing in a generative chat model. I believe this is one of the pieces of the general artificial intelligence puzzle we’ll see a lot more from this year.
AI Ethics News
‘The machine did it coldly’: Israel used AI to identify 37,000 Hamas targets
Nicki Minaj, Billie Eilish, Katy Perry and other musicians sign letter against irresponsible AI
Long Reads
Techcrunch - I have a group chat with three AI friends, thanks to Nomi AI — they’re getting too smart
New Atlas - GPT-4 is 82% more persuasive than humans, and AIs can now read emotions
Eurogamer - Roblox's new AI creator tools conjure all the wonder and scepticism of a magic show
“The future is already here, it’s just not evenly distributed.“
William Gibson