Welcome, humans.

Exciting news: Last Friday, our new podcast The Neuron: AI Explained became the #1 Technology show on Apple Podcasts! The rave reviews are pouring in.

Be sure to tune in, enter our $500 Visa gift card giveaway by subscribing and sharing, and send us your thoughts on the pod (& on Pete’s background plant)—we read every comment!

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

• People are livid over Instagram’s new AI search bar.

• Meta’s Ray-Ban smart glasses are turning heads.

• Apple and OpenAI are discussing integrating OpenAI tech into iPhones.

• Sam Altman said that GPT-6 will be way smarter than GPT-5, which will be way smarter than GPT-4.

On Saturday’s podcast: Elon Musk’s xAI raising $6B (!), Worldcoin’s OpenAI partnership, and Synthesia’s lifelike avatars (Apple Podcasts, Spotify, YouTube).

April was a really strong month for Meta AI…until it wasn’t.

On 4/19, Meta announced its latest Llama 3 models, the most powerful of which will be the first open-source model to compete with ChatGPT-4. Most notably, Meta plans to seep its Meta AI chatbot into all of its products: WhatsApp, Insta, Facebook, and Messenger.

• Investors weren't happy—$META plummeted as much as 19% after hours on Wednesday after Zuck announced increasing investments in AI infrastructure.

  • Also trending: Microsoft, Meta, and Google spent $32B+ on data centers in Q1.

• Users weren’t happy—people DID NOT dig the new AI features that they’re seeing in their (anti)social media feeds.

Specifically, there was fury around the new “Ask Meta AI anything” search bar that’s now the default option when you want to search for a user or video.

Yeah…it’s a pretty horrendous feature.

1. It actually makes it way harder to find what you’re looking for (just try looking up a friend’s account).

2. Its AI suggestions aren’t at all personalized (it should know better than suggesting cardio workouts to us), which is shocking given how much data Meta has on us.

3. It can’t be turned off, which is just incredibly frustrating.

Meta AI was also caught doing super bizarre things, like telling an FB group that it has an academically gifted and challenged child attending NYC public schools. WTF!

But it's not all doom and gloom for Meta AI—they’ve rolled out a genuinely useful tool that, IMO, deserves more spotlight (details to follow).

FROM OUR PARTNERS

Pair AI with your datastore using this new AI tool!

ChatGPT-4 is nice, but businesses that integrate their business data—CRMs, spreadsheets, internal policies—with LLMs will see the highest ROI from AI. 

OctoStack from OctoAI helps you harness advanced AI models to your data safely and securely.

And they’re hosting a webinar on May 7th to show you how to:

• Set up data systems and processes on your Snowflake platform using OctoStack.

• Use Retrieval Augmented Generation to make your AI outputs more reliable.

• Ensure your AI integrates smoothly with your data systems using JSON.

Register here and join for the technical Q&A. Everyone who signs up will receive a recording.

Meta’s AI-powered Ray-Ban glasses should be given a closer look.

When it comes to AI gadgets, everything in the media is “Humane-this” or “rabbit-that”. We suggest taking a closer look at Meta’s Ray-Ban smart glasses—check them out here.

Here’s the scoop: the $300, optionally prescription Ray-Bans are glasses that “see” what you see. They’re multimodal and pair with Meta AI, which can interpret what you see and interact with your phone.

It’s like Siri meets ChatGPT meets your immediate surroundings.

For instance, you could say…

• “Hey Meta, what kind of tree is that?”

• “Hey Meta, send a voice message to my Dad about my job update.”

• “Hey Meta, play my favorite Spotify playlist.”

For us, the best part about these glasses is that whenever you want to snap a photo or video, just tell Meta AI to handle it, skipping the hassle of pulling out your iPhone.

Here is the best review we’ve seen yet:

Why it matters: There’s AI that’s actually useful (like ChatGPT), and then there are the “AI-powered” features that big tech companies ship to signal to the markets that they’re being innovative in AI.

We’ll call out the BS wherever we see it while also highlighting real-world use cases of AI that we find genuinely useful!

Around the Horn.

• Apple and OpenAI have resumed talks about incorporating OpenAI’s tech in upcoming AI features for iPhones.

• Sam Altman: “we can say right now, with a high degree of scientific certainty, that GPT5 is going to be a lot smarter than GPT4, GPT6 is going to be a lot smarter than GPT5 and we are not going to get off this curve”.

• Guide for Beginners: Custom Instructions and Memory explained and compared.

Treats To Try.

1. *Cutting-edge businesses are turning to synthetic data to train better AI models. Learn how to do the same by downloading Gretel’s free Definitive Guide to Synthetic Data for the Enterprise!

2. Notta translates your videos into 15 languages while keeping your natural voice the same.

3. Patterns is an AI-powered data analyst that integrates with your databases, catalogs, semantic layer, and business intelligence.

4. LangWatch provides tools for improving AI models through real-time analytics and safety features

5. Academic AI is a GPT that lets you conduct research from 300M publications.

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

Monday Meme.

A Cat's Commentary.