MSBuild - Microsoft has been BUILDING (and a bunch more stuff. TFG006)

"We are nowhere near the point of diminishing marginal returns on how powerful we can make AI models as we increase the scale of compute" - Kevin Scott, CTO 微软 . Last week we saw OpenAI steal 谷歌 's thunder on Monday, Google proceeding to reveal the same thing as OpenAI the next day, and now - a week later - we are hearing from Microsoft, and they mean business.

Many reading this know that Microsoft recently launched a division - Microsoft AI Co-Innovation Lab - and poached two cofounding executives from Inflection AI , Karén Simonyan and Moustafa Suleyman (cofounder of DeepMind previously), Reid Hoffman's "right and left hand man" in the company. Microsoft bought LinkedIn from Reid and team almost a decade ago for $26B, and Reid was originally on OpenAI's board, is presently on Microsoft's board, and was (via Inflection) compensated $560M just over a week ago when Karan and Moustafa were tapped to launch the division (effectively an acquisition to satiate investors in Inflection) as Microsoft gained access to all Inflection's technology.

Here's where it gets really interesting:

The tech that Inflection AI built, Pi (personal intelligence) is basically what both OpenAi and then Google announced last week. "Emotionally intelligent" "Her-style" (see: law suit: Scarlett Johansen and OpenAI) potentially a-bit-too-flirtty voice bots, letting you speak with both ChatGPT and Gemini (I've been doing that for months, but this does take it to the next level, and is now actually "multimodal", which is really cool. Microsoft barely even mentioned this tech, which they'll definitely be rolling out too as they're the ones who know own the tech, whereas it was OpenAI and Google's focal points of their"big news" last week. I do have word that OpenAI was ready to reveal more stuff last week if Google did, but Google is usually a bit slower to show their hand, and so here we are.

But what Microsoft is revealing at Build is HUGE. They brought out Sam Altman. they brought out Sal Khan of Khan Academy , and Kevin - Microsoft CTO who I quoted above - had some pretty incredible things to say and show. Copilots (and Google's equivalent as well as ChatGPT-o's) are going to turn any desktop and then device into a no-code developer suite, agency, team of assistants working autonomously and much more. Multimodality is a big deal because of how it inputs and outputs, sight and vision play into this (there are already AI powered robot teachers in both Texas and India), and partnering with companies like Khan Academy who's vision and goal is "ubiquitous higher education to anyone", and hearing Kevin say how these systems are about to get SO much better SO soon and then exponentially from there is incredibly invigorating. Also notably, well-funded 'Devin' and its company, Cognition Labs, turns out not to be "bogus" at all, despite concerns when it seemed some months back that it's demos were faked, and is partnering with Microsoft in a way that sounds legitimately game-changing.

Like, actually game changing.

But to shift gears: Cognition Labs aren't the only ones who've recently faked demos that I've written about here that turn out to be up to some really cool soon too: notably, Tesla's robot demo folding clothes was determined to be fake, and I've personally been saying for a week that I believe that OpenAI's demo last week of "omni" was likely faked (I that I can prove it, and have), and potentially Google's the next day as well, but the reality is that these companies have REAL and really incredible tech they've already built, tested, and are ready to deploy as soon as humanity is ready.

You have likely seen recent posts of mine linking Microsoft to 苹果 , to AMD , to GameStop , and LinkedIn , AMD and 英伟达 , and then linking Google to 三星电子 and so many more interconnectivity's, including airlines, vehicle manufacturers, and so much more, but effectively all companies showing their hands and current publicly facing iterations of AI are starting to say the same things, and I believe they're worth listening too. Namely:

• AI will now finally start learning from itself and getting smarter and smarter
• the next powerful step-up in what's possible with AI is about to be unleashed and it's a true step up (level-up, next-gen, et al)
• AI now is multimodal - can process different types of inputs and outputs simultaneously, and can listen, see, talk, create
• Autonomous Agents are coming in hot - this is when AI can take control of your computer (soon phone) and do things on your behalf, or create needed personas that can "work" with each other to complete tasks and objectives on your, the users', behalf
• today's AI is "garbage" compared to what we'll soon have
• AI can now both code on its own and help programmers exponentially in ways that most have said isn't possible for a long time to come
• AGI is closer than we think. My assessment is that when we combine robots with Large Language Models, Large Actions Models, Large Nature Models, Large World Models and several others (I'm personally working one one as "we speak") we are effectively "there"
• Copilots and similar like Google's version that can tie into all your files, histories, etc and act as consultants, assistants, data analysts, programmers, financial experts and more will change the way we humans live our lives
• Avatars, "brains" (check out Intel's new project, Hala Point), teachers, therapists and more will all be en fuego soon. The only way forward is to learn all this stuff NOW
• We do NEED more compute. Sam Altman wasn't wrong although he was unsuccessful in raising $7T (seven trillion dollars) for it
• Quantum computing and super computing are coming, as well as newer, better chips, and compute is getting faster, cheaper, and attainable

This is definitely the time to be paying attention, learning, adopting, and understanding all the changes currently at hand. You're taking a big step in reading and engaging with this newsletter. If you find it beneficial, please share it with some friends, share this issue, and reach out with feedback.

One of the goals of this newsletter, fully titled "Tech For Good: Building In Public Using AI" is to - as the name boldly proclaims - teach how to use AI to build stuff, and to build stuff while allowing others to "watch over our shoulders". To that end, my startup WeJ (Where Everyone's Jamming - disrupting the music space as a community using blockchain and g-AI to empower DJs worldwide) have been dedicating 20 mins of our weekly team standups to a live stream on social medias, and we will further open those calls by sharing the zoom links here, on my profile, and elsewhere publicly to allow anyone to hear what we're building, how, what AIs we're using, how, and how you can leverage the same in your ideas, businesses and lives too. Stay tuned for more on that, but it's an idea that's been percolating, and now that we've been accepted into the Story Protocol, are in the process of onboarding some globally known, world class DJs and producers, have partnered with Blockbook and more - and we're eager to share what, how, and all that good stuff with our readers of Tech For Good, as well as with the world as a whole. As I love to say, TOGETHER WE RISE!

(If you're interested in gaining access to us building public please comment or DM me.)

I loved my old AI newsletter that had global reach and nearly 30 issues, "AI News" particularly because every week I'd showcase a prompt of the week and a tutorial of the week. In the spirit of building in public, I'll begin baking some of those back into Tech For Good. That said, I encourage those new to the newsletter to seek out the first issue and two of this, where I link to the previous newsletter and teach "how to use it" using AI - the nearly 30 prompts and tutorials span so many categories, and are legitimate growth hacks and ways to stay on top of all things AI, how to use it, use it well, and know how to use and leverage all tools coming down the pike. YOUR FEEDBACK IS MORE THAN APPRECIATED - ITS MISSION CRITICAL. Please provide feedback on what I can discuss, what you'd like to learn how to do and use AI for, and how I can make this more compelling. Right now I'm doing this as a labor of love but will likely hire a team as well to include great visuals/graphics, interactivity, and more but I don't want to make those moves without YOUR feedback and input.

In other news, now music related as that's where I'm currently spending most of my time and attention, I recently shared that Eleven Labs is getting into the generative AI music space, and have some INCREDIBLE capabilities, but it now seems that even more recently Suno AI has upgraded their tech, and are competitively incredible. Audio also has great music generating tech, but in my opinion Suno right now is the one I'm most blown away by. In the spirit of "building in public" I will share a recent track I produced (PLEASE listen with good speakers/headphones) here https://suno.com/song/a378acb0-8351-4b21-aa1b-c442b8aa7a0a - I wish I could share the prompt and what platform generated it, as I don't believe that it was Suno itself (although they do generate great lyrics), but I've been testing all platforms so much in the past week that I can find the prompts, platforms, etc. I need the "recall" feature that Microsoft AI just revealed and then I'll be able to find, no problem. What I can find is that for "style of music" I used "trap afrobeats heavy bass" and the lyrics (again, I believe generated elsewhere, but prompted by me):

[Verse]

I'm on a love trip, it's Dr Sha's prescription

Mixing trap beats with some Afrobeats addiction

Sipping on the potion, it's a sick composition

Got that fire flow, causing widespread ignition

I'm the trap doctor, here to heal your heartache

Got the afro vibes, making bodies shake

With the heavy bass, we gon' make it quake

Dr Sha's in the house, it's time to celebrate

[Verse 2]

My rhymes are sharp, like a surgeon's precision

Cutting through the beats with my verbal incision

I'm the love doctor, prescribing sweet affection

Got that trap flavor, causing mass infection

LOLOLOLOLOLOLOLOLOLOLOL better than some of the lyrics in the song I shared from Eleven Labs (I wish the "link" function/feature was working here on LinkedIn newsletter, like the @ tag feature, but I'm sure that's coming).

But I don't recall or can't currently find the prompt I used to generate those - as those aren't lyrics I wrote. Typically I write my own lyrics when using Suno, Udio, etc as I myself am a lyricist, freestyle rapper, battle rapper (winner), creative, etc. but I do find that when AI writes the lyrics they're often more rhythmic, "real" sounding, and optimized (at least not yet).

I HIGHLY, HIGHLY, HIGHLY recommend checking out https://suno.com. Please share your best generations in the comments.

I used to feature 4 products of the week in my proviso global newsletter, "AI News", and will start featuring products here soon, primarily on merit and what truly impresses me but I do appreciate the numerous inquires about being promoted in this 130K+ subscriber newsletter (in under 6 weeks). I'll also keep talking about blockchain and promoting amazing platforms and protocols like Story Protocol, Blockbook, Decentralized Labs, Data Profit, Polygon, Header Hashgraf, as well as quantum computing, robots, life sciences, investment, entrepreneurship and much more. I also am keenly aware of the AI that LinkedIn is building on their own, in tandem and with support from Microsoft, and how to leverage that too. More in future issues. There's also so much more that I want to include thusly, so will commit to another issue this week.

THANK YOU for being an early subscriber and reader of Tech For Good - I hope it adds perspective and insight actionably into your life!

#techforgood #ai #ainews #blockchain #web3 #tech #innovation #microsoft #linkedin #google #metaai #coryconnects #tech4good #agi #msbuild #copilots #autonomousagents #multimodal #openai #chatgpt5 #sama #corywarfield #coryconnects #linkedinnewsletters #linkedinvideo #inflectionai $litrendingtopics #topvoices #staytuned