😺 AI geniuses win Nobel Prize

PLUS: OpenAI owes Microsoft HOW much?!

Welcome, humans.

Did you know we now offer an audio version of the newsletter? Click “Listen Online” at the top of this email to enjoy your favorite newsletter on the go (or sitting down!).

Now listen to this: Microsoft’s profit share stake in OpenAI is WILD—the big M gets 75% of ALL of OpenAI’s future profits, until the $13B it originally invested gets repaid. This means OpenAI has to generate $17.3B in net income before it can generate its own profits. As Jack Raines pointed out, only 17 companies made $17B last year…

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

Two AI legends won the Nobel Prize in Physics.
OpenAI sidestepped Microsoft to secure more datacenter power.
Amazon built an AI advisor with internal HR data.
New video tool can generate footage of you speaking anywhere.

Advertise in The Neuron here.

Two AI legends (and one famous AI “Doomer”) just won the Nobel Prize.

Here’s the live announcement video and a breakdown of their reasoning.

Did AI just peak?! First, OpenAI raised the largest VC round in history, and now, two Godfathers of AI just won the Nobel Prize in Physics.

You read that right—”Physics”, not “Math” or “Computer Science” (there actually isn’t a Computer Nobel Prize, FYI). Cold. Hard. Physics.

Here’s why: John J. Hopfield and Geoffrey E. Hinton, developed the foundations for machine learning that all modern AI relies on.

While deserved, this was a controversial move. Some called out the Nobel committee for giving the 2024 Nobel Prize in Physics to “not physics.”

The Nobel Prize folks justified their decision by pointing out how neural networks are used in “particle physics, material science, and astrophysics.”

The history of machine learning (in brief):

1982: John Hopfield creates the “Hopfield network”, an AI system that mirrors the human brain with artificial "neurons" that can recall patterns like how we remember words and concepts.

1983-1985: Geoff Hinton creates the “Boltzmann machine”, which uses physics to identify common traits in large data sets, allowing it to recognize and classify images… or even… generate new ones.

A year ago: After a decade working at Google, Hinton quits his job to “talk about the dangers of AI.” That, and he was 75 and ready to retire. Just saying.

FROM OUR PARTNERS

The future of AI is local, ownable, and in your hands—thanks to this “privacy-first” AI…

Powerful. Effortless. Private. Doesn’t that sound nice?

That’s webAI.

You don’t want to give all your data to big AI cloud providers. And you want more control over your data and models, too. That’s why webAI is your new BFF.

With webAI, you can: 

Test and iterate solutions directly, with no need for a dedicated AI team.
Utilize local compute, meaning faster, secure testing (at lower costs).
Deploy effortlessly across local, cloud, edge, or hybrid environments.

Imagine having the power of cutting-edge AI right at your fingertips, without the cloud costs, data privacy concerns, or the need for a Ph.D. in machine learning on call.

You can unlock the real value of AI with local, ownable, customizable models that YOU control. Click here to learn more.

Around the Horn.

More cool AI models from Meta in this thread

OpenAI is worried Microsoft isn’t moving fast enough to supply it with enough computing power, so its planning to rent additional servers from Oracle at a datacenter in Texas that could power multiple big cities.
Uber will release an AI assistant to “help drivers transition” to EVs.
Amazon built its own genAI “culture coach” trained on performance reviews and promotion docs, that can then give employees advice to “feel stronger.”
Pieter Levels, founder of NomadList created a “speaking AI video” that lets you “put yourself anywhere in the world speaking about anything you want.”

Treats To Try.

*Craving NVIDIA H100 performance, but NOT the price tag? DigitalOcean's GPU Droplets offer on-demand, virtualized GPUs for flexible, affordable AI development. On-demand H100 power, easy setup, and fast deployment: what more do you want? Get Started Now.
Hailuo AI, the company behind video model MiniMax, released an image to video model you can try here.
Eddy AI is a new free tool to quickly edit interview footage with text prompts. We tried it and made a rough cut in ~30 seconds.
Cove is a collaborative workspace to explore ideas, plan projects, and work alongside an AI that interacts with your content and learns from you (raised $6M from Sequoia).
Kvistly creates quizzes for learning, remote training and team-building events (new team, just launched with a free demo).
Unify, an OpenAI Converge accelerator grad, sends personalized messages to potential buyers at the right time (raised $12M).
Anthropic launched “Message Batches API” that lets you send up to 10K queries at a time at 50% of the cost of a standard API call (they get processed in “less than 24 hours”, FYI).

See our top 51 AI Tools for Business here!

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

Where do you #Neuron?!

Submit where you #Neuron here for a chance to be featured in our newsletter next week!

Hayden from Asheville, NC: “Our little community help center after the storms in Asheville. Reading the Neuron via Starlink internet because we don’t have cell service yet!”

(Wow, thank you for taking the time to read OUR newsletter, even with everything you’re going through!)

P.S: If anyone out there wants to help out w/ the Hurricane Helene recovery, see this thread or donate directly here.

Matt via Amtrak train from Philly to NYC: “90 minute train ride is the PERFECT time for me to expand my AI knowledge and experiment with new tools that The Neuron recommends!”