😺 Mind-blowing ChatGPT-4o demos

PLUS: A 48% increase in greenhouse gas emissions 😷

Welcome, humans.

We’re working on some seriously fire new educational content for The Neuron that we think you’ll love. It’s like Tom & Jerry meets AI explainers, except our cat always nabs the pesky house mice you've been chasing.

Here's a sneak peek at a situation you've all been guilty of by now:

Here’s what you need to know about AI today:

  • We’re spotlighting some incredible demos of ChatGPT-4o with vision.

  • Runway is in talks to raise $450M at a $4B valuation.

  • Google’s carbon footprint has surged due to more data centers needed to power AI.

  • Perplexity rolled out a feature that provides more thorough and better-researched responses.

What happens when ChatGPT can see everything you can…

Some folks daydream about the Lambos they'll buy when they hit it big. Me? I’m daydreaming about the epic stuff I’ll pull off with ChatGPT-4 when its vision and Voice Mode features are finally here.

The release date on these is TBD. It was supposed to be June, but OpenAI's doing more safety testing. Probably this month, but who knows with these AI magicians.

Anyway, we're drooling over this model even more this week, thanks to some sick demos the team's been dropping.

First up, the voice stuff from OpenAI's YouTube channel:

  • Our favorite: creating different character voices with GPT-4 voice (link).

  • Roleplaying an interview with GPT-4 that can 'see' the interviewer (link).

  • Using ChatGPT as a language tutor to practice your Portuguese (link).

These demos are unlike anything we've seen before.

But the vision stuff is what's really got us buzzing. OpenAI's developer advocate Romain Huet showed off some of these capabilities at the AI Engineer's World Fair last week, and holy moly:

  • It summarized a page from a physical book (yeah, those still exist) in a second flat just by glancing at it (link).

  • It understood sketches of images—your kindergarten art skills might finally be useful! (link)

  • Most importantly: it helped Romain fix a wonky design by looking at his screen and tweaking the code on-the-fly (link):

This last point is why it matters. Soon, we might have an AI that sees what we see on our computer screen 24/7. No more struggling to describe what you’re seeing or sending a million screenshots. It'll be a real-time AI assistant for all your visual tasks.

That is…if they ever release it. Come on, OpenAI. We're not getting any younger here!


How to build AI applications that aren’t inaccurate.

Raise your hand if this sounds familiar: you expect AI to handle enterprise tasks, but it lacks domain expertise and live info, making its answers inaccurate.

Sure, you could fine-tune the models, but that’s expensive, labor-intensive, and needs technical know-how.

Instead, try NVIDIA AI Enterprise on AWS (check out this webinar to learn how).

NeMo Retriever, part of NVIDIA AI Enterprise on AWS, helps companies develop AI apps with RAG (retrieval-augmented generation, aka AI that uses your data for accurate responses).

Around the Horn.

  • Runway, an AI video generator, is in talks to raise $450M at a $4B valuation.

  • Figma hit pause on a feature that creates designs from text prompts after it churned out a weather app eerily similar to Apple’s.

  • Google’s greenhouse gas emissions have risen 48% in the last 5 years thanks to the AI-fueled data center boom.

Treats To Try.

  1. *Neighborhood Sun is a cleantech startup helping to connect everyone to affordable and accessible solar energy. Invest in Neighborhood Sun on Wefunder and become part of a sustainable future.

  2. Meta’s 3D Gen is new research that can generate 3D assets from text prompts in less than one minute.

  3. Resemble AI lets you clone your voice AND has a tool that sniffs out audio deepfakes with 94% accuracy.

  4. Proteus breathes life into still images, making snapshots of you or others laugh, rap, sing, and more.

*This is sponsored content. Advertise in The Neuron here.

A Cat's Commentary.

Where do you #Neuron?!

Thanks to everyone who submitted a pic of where they read The Neuron. Let’s just say WE. ARE. SO. BACK.

Submit where you #Neuron here for a chance to be featured in The Neuron!

Pep (the cat's name is Naia), from Catalonia: I receive different newsletters about AI but this one is the best one!

Cynthia (from Southern California): I need to keep learning more about AI…!

Eve: It’s not just a matter of where we Neuron, it’s also with whom!"

Michelle: From the nation's capital to the Great Plains, the Neuron AI news travels with us. (PS. I am not the driver).

Andrew (from Hockessin, Deleware): Before starting summer school, I always get grounded by reading The Neuron Newsletter. Keeps me one step ahead of these kids!

That’s all for today, for more AI treats, check out our website.

Get your brand in front of 450,000+ professionals here.

See you cool cats on Twitter: @nonmayorpete & @noahedelman02

What'd you think of today's email?

Login or Subscribe to participate in polls.