• Everyday AI
  • Posts
  • Microsoft Build AI Recap - Everything you need to know 🛠️

Microsoft Build AI Recap - Everything you need to know 🛠️

Copilot updates and on-device AI, Amazon to launch conversational Alexa, Humane looking to sell off its Ai Pin and more!

Author

Everyday AI
May 22, 2024

Outsmart The Future

Sup y’all 👋

We’re letting you all take control again and help us decide what we should talk about on tomorrow’s show!

Which of the options below do you wanna hear more about?

What should we cover on tomorrow's show?

Login or Subscribe to participate in polls.

✌️
Jordan

Today in Everyday AI
7 minute read

🎙 Daily Podcast Episode: Microsoft has unleashed a slew of new AI updates and devices. So what does it mean for the competition and all of us? We break down the 5 AI announcements you need to know about Microsoft’s Build conference. Give it a listen.

🕵️‍♂️ Fresh Finds: Microsoft CEO talks Copilot + AI PCs vs Macs, Meta’s AI chief says LLMs won’t reach human intelligence and TikTok adds GenAI features to its ad platform. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: Amazon to launch conversational Alexa for a monthly fee, Humane looking to sell off its AI Pin and Meta’s new AI model and the FCC might be reigning in AI even more for political ads. For that and more, read on for Byte Sized News.

🚀 AI In 5: We’re breaking down Microsoft Build’s slew of AI announcements and giving you 5 things you need to know in (about) 5 minutes. See it here

🧠 Learn & Leveraging AI: If you’re wondering why Microsoft’s recent AI moves are so crucial to the GenAI world, we’re breaking down what you need to know. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Google calling out Microsoft's security, EU AI Act to go live and Scarlett Johansson calling out OpenAI. Check it here!

Microsoft Build AI Recap - 5 things you need to know 🛠️

To end a week-ish full of AI happenings, Microsoft has thrown all kinds of monkey wrenches into the GenAI race.

What did they announce at their Microsoft Build conference?

And how might it impact you?

Our last takeaway may surprise you.

Join the conversation and ask Jordan questions on Microsoft AI here.

Also on the pod today:

• Microsoft Build Conference Key AI Features 🔑
• Microsoft Copilot Updates 💻
• On-device AI and its future 📲

It’ll be worth your 50 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Findr is an AI search assistant for work, CANVID is an AI screen recorder and Nowadays is an AI assistant for corporate event planning.

Trending in AI – In this exclusive interview, Microsoft’s CEO discusses how AI Copilot + PCs can beat Macs.

Big Tech - Meta’s AI Chief says that LLMs won’t reach human intelligence.

LLMs - New features just dropped on Google’s Gemini advanced, as it now supports spreadsheet uploads to help you better analyze your data.

Money in AI – DeepL, a startup specializing in AI language translation, has secured $300 million in funding.

Social Media – TikTok is boosting its ad platform with new GenAI features.

AI in Politics — The U.S. House of Representatives is holding a hearing on security and AI.

1. Amazon to Launch Conversational Alexa with Monthly Subscription 🗣

Amazon is reportedly planning to launch a more advanced conversational version of Alexa later this year, aiming to compete with new generative AI-powered chatbots like Google’s Gemini. The move follows recent advancements in artificial intelligence, with OpenAI unveiling GPT-4o capable of more in-depth two-way conversations than Alexa.

This shift to a subscription-based model for Alexa suggests a significant evolution in Amazon’s voice assistant strategy, potentially revolutionizing AI interactions for users.

2. Tech Startup Humane Seeks Buyer for AI Pin Wearable 📍

Humane is actively looking to sell its AI Pin wearable computer, despite receiving poor reviews for its slow performance and underwhelming user experience. Priced between $750 million and $1 billion, the sale could attract interest from industry behemoths like Amazon, Apple, Google, Meta, and Microsoft who are all heavily invested in AI technologies.

With backing from notable figures like OpenAI CEO Sam Altman, the potential acquisition raises questions about the future of wearable AI devices.

3. Meta's Chameleon AI Model Revolutionizes Multimodal Tasks 🦎

Meta's new Chameleon AI model excels in various tasks involving both text and images, outperforming other models in visual question answering and image captioning. The Chameleon model uses an innovative early-fusion token-based approach to seamlessly process and generate sequences of both images and text, unlocking new capabilities in mixed-modal reasoning and content generation.

Researchers trained the Chameleon model using 44 trillion tokens, achieving state-of-the-art performance through meticulous training on Nvidia A100 80GB GPUs for over 5 million hours.  

4. Anthropic Unveils Revolutionary Insights into AI Models 💡

A groundbreaking peek inside Claude Sonnet has unearthed millions of internal concepts, shedding light on the inner workings of large language models. This dive into interpretability could pave the way for safer AI models, offering a glimpse into how these models process information and make decisions.

The discovery challenges the black box perception of AI, revealing a complex network of features representing diverse concepts from scientific fields to city names.

5. NVIDIA's Earnings Surge Drives US Stock Market to New Heights 📈

NVIDIA's impressive 90% surge in share value this year places it third in U.S. market cap, just behind Microsoft and Apple. With NVIDIA being a leader in AI chips, its performance serves as a crucial indicator of the AI industry's growth and investor enthusiasm. The company's significant weight in major indexes like S&P 500 and Nasdaq 100 amplifies its impact on the broader market trends.

6. FCC Proposes AI Transparency in Political Ads 🤖

The Federal Communications Commission is pushing for transparency in political ads by requiring disclosure when AI-generated content is used on broadcast television and radio, as reported by the Associated Press.

FCC chair Jessica Rosenworcel aims to have regulations in place before the upcoming U.S. election to combat the potential misuse of AI tools in political communications.

The proposal would mandate broadcasters to verify if AI tools were utilized in creating political ad content, addressing concerns about misleading voters through AI-generated images, videos, and audio clips. This move reflects a growing need to regulate AI in political advertising and enhance transparency in the digital age.

Microsoft Build Recap - 5 New AI Features You MUST Know!

Microsoft released so many new AI features and products at Build 2024 that it’s so hard to keep up with.

That’s why we’re giving you the breakdown of everything that was announced and the 5 biggest things you need to know about Build 2024.

🦾How You Can Leverage:

Is there a new AI sheriff in town? 🤠

The PC maker is about midway through its 3-day Microsoft Build conference in Seattle, and they’ve been dropping AI announcements nonstop like trying to catch a water balloon while drenched in canola oil. 

After Google and OpenAI went blow-for-blow with new AI announcements last week, it looked like Microsoft might softly ride off into the sunset without much fanfare. 

Nope. 

From new products and services, brand new versions of Copilot and even AI agents, Microsoft is feeding all us AI aficionados the digital dopamine we need. 

Head spinning yet? 

Over the course of about 9 days, we’ve seen literal game-changing AI announcements from the 3 of the biggest players in the game — OpenAI, Google and now Microsoft. 

If you haven’t been able to dive deep to see what it all means, we got you. 

Today, we dished on the 5 things that really stood out to us about Microsoft’s latest AI plays. (The last one may surprise you.) 

Go give today’s episode a watch/read/listen, and then let’s dive a bit deeper on what you need to know. 

Let’s gooooooo 👇

5 – Microsoft Edge AI goes mainstream 😎

Here’s the real talk — Edge AI is a big deal. 

Bringing the power of Generative AI and Large Language Models to a local device is undoubtedly the future of the space. 

(More hot takes on that later?) 

Edge AI, or on-device AI is faster. (Technically)
More secure.
And more connected.

In simple terms, right now, most of our GenAI work pings to the cloud then back to us, causing quite a bit of latency. 

When devices run AI locally — light speed. 

The technical speak here:

  • The new Copilot Plus PCs will feature Snapdragon X Elite and Plus processors from Qualcomm, capable of 40 trillion operations per second, and neural processing units designed to power new AI features in Windows 11. 

  • And to run LLMs like GPT-4o locally on laptops, Microsoft is introducing a new hardware component called the Neural Processing Unit (NPU), designed specifically for running generative AI models like GPT-4o locally on PCs and laptops.

So, it’s a combination of traditional CPUs and GPUs with these new NPUs that will reportedly make this all possible. 

What it means: 

Microsoft is making a big splash, and it could be risky. Working with local models is a huge paradigm shift.

It opens up a new way to work, and potentially seeing more and more employees accessing private and secure company docs on their local machines vs. secured cloud storage. 

Will the trade-off be worth it? 

Probably. 

But, this opens up a whole new host of security features that companies will need to address. 

4 – Recall = AI powered memory 🧠

You ever spend like 52 minutes looking for ONE piece of information from a meeting from last week? 

(Just us?) 

Think of Recall like Internet history in your browser, but for your entire computer.

Throw in the ability for semantic search, and you’ve got a natural language AI companion that can quickly rewind through your entire PC activity to find that golden nugget you were scouring the earth for. 

What it means: 

A new way of work. For real. 

Looking into the future a bit, we see the average employee who uses Recall will likely change how they work.

It could go one of two ways —

  1. The average worker may kinda check out and over-rely on Copilot AI and Recall.

  2. Or, the average employee becomes a legit superhuman and can climb the company ladder quicker than those firefighter racing videos. 

3 –A more powerful Copilot that gets you ⚡️

Combine #1 and #2 above, add that into the ‘standard’ Microsoft Copilot product, and you can see how the new and improved Copilot is gonna understand you a bit better. 

And better yet? 

Microsoft integrating OpenAI’s GPT-4o into Copilot. The ability for Copilot to instantly ‘see’ anything.

Live.
In real-time.

And then for it to be able to perform tasks for you?! (See below for more on that.) 

That’s where the future of work really starts to get rattled. 

What it means: 

Imagine having these kinds of conversations multiple times an hour to finish your work.

We see the actual interaction with computers drastically changing in the coming months as Copilot’s updates (and GPT-4o voice updates) roll out. 

Talking to your computer and then WATCHING it do your work is gonna be the norm.

Wild times we’re living in, right?

P.S. — Like we talked about on today’s show, overbearing companies who tend to micromanage could likely use this feature against you. So if you’re one of those crafty remote who bought a mouse mover/jiggler to ‘look active’, this new Recall feature could in theory bite you in the butt if companies are asking for a Recall report.

2 –Microsoft could make agents mainstream 🕵

Sure, Meta and Google have already announced AI agents. 

Meta snuck in a reference to building agents in its Llama 3 announcement, but we’ve yet to see anything come to fruition. And Google hyped up its Gemini-powered agents, which could still be months/quarters away. (Plus, we know how Google is at times more marketing than reality.) 

Microsoft could be first to market, though, as they announced a gradual 2024 rollout for building Agents in Copilot Studio. 

What it means: 

Instructing your AI assistant on what to do is so…. 2023? 

  • Agents can act independently to automate complex business processes.

  • These Copilot agents can be triggered by events (not just conversations or manual prompts), understand intent, access data, check inventory, respond to users, and complete tasks over days.

  • They can be tailored for various roles in industries like IT, marketing, sales, and customer service.

  • Copilot agents use generative AI for planning, reasoning, memory, and can ask for help when needed.

Go watch this quick 9 minute video that goes over Copilot capabilities. 

Click to watch the 9-minute AI agent recap

1 – Microsoft is fighting the wrong battle 🤦

#HotTakeWednesday? 

Here’s our take on this: Microsoft is barking up the wrong tree. 

It seemed like a huge focus for Microsoft was promoting how much more powerful its new Surface Copilot Plus laptop was vs. Apple’s MacBook Air M3. 

Cool? 

TBH, we’re not sure why Microsoft is kinda obsessed about cutting into the MacBook market share. And we get it, Microsoft is looking to sell 50 million PCs in the next year.

Chances are, though, Apple may one-up the performance of Microsoft’s latest surface laptop when it releases its M4-powered Macs later this year. 

What it means: 

Cold hard facts — if Microsoft wants to sell 50 million PCs this year and focus on eating into Apple’s slice of the pie, they should stop focusing so much attention on speed and performance benchmarks. 

Those are gonna change. 

Instead? 

They should focus more attention on creating in-roads for die-hard Mac fans to make it easier for them to untether from the tight grips of the Apple ecosystem. (Aside from simply support Parallels.) Kinda like how Google did with free and unlimited storage for iPhone users in Google Photos. 

Microsoft needs an Apple Trojan Horse, not another ‘head to head’ specs battle.

⌚

Numbers to watch

65%

According to a Salesforce study, 65% of global C-suite executives believe trust in AI drives revenue.

Reply

or to participate.