The Week In AI: Cheap House Robots and Pricey Manhattan Projects

Last Friday, my colleague Ethan Feller and I had another awesome live event surveying all the latest happenings in AI.

We do so every Friday in an X Space where we each have 6 unique AI topics that we present in a rapid-fire 60-minute run-through.

We started this gig in May when the innovations, controversy, and drama were happening so fast, we didn't see how it could continue at that pace. Eight weeks later, it only shows signs of more acceleration and amazement.

So here's a quick summary of just 6 of the major reveals, deals, and feels we covered last Friday July 11...

1) $9,000 Humanoid Robots from K-Scale Labs in Palo Alto

K-Scale Labs launched pre-orders for its open-source humanoid, priced at $9K. The K-Bot stands 4′7″, weighs 77lbs, and integrates with K-Scale's open-source stack, covering various models and hardware designs.

One of the founding engineers, working out of their shared house in Palo Alto, posted this on X: 

Robotics in the U.S. is broken -- proprietary, expensive, and slow. 

We’re going to change that. We're building, learning, and shipping mass-production grade humanoid robots with the open-source communities.

We're launching America's first open-source end-to-end humanoid robot starting at $8,999. Open-source hardware, Rust OS, Python SDK, RL training libraries.

The future of robotics is autonomous, open-source, affordable, and made in America.

(end of X post by @JingxiangMo)

Getting a Jump on Optimus and Figure

Even though we talked about these renegades in an early episode of The Week In AI, I was newly invigorated and inspired by their passionate 3-minute video that chronicled their journey and mission -- as they just released 100 new models for sale and raised $1 million dollars very quickly.

Essentially, the K-Scale team, led by a former developer at Tesla TSLA and OpenAI named Benjamin Bolte, wants to beat both Tesla Optimus and Figure AI robots to the market with an open-source platform.

I am humbled and awed by their vision and tenacity when they could easily try a smoother path with more investment funding to scale up instead of going direct with "models in progress." 

But then again, isn't that what the big LLM model companies are doing?

Benjamin Bolte thinks we should all be able to train our humanoid robots to do a specific task -- mop the floor, wipe the counters, vacuum, load the dishwasher, or fold the laundry -- and that homeowners can share those trainings across the open-source "robo-sphere" to help everyone have a better experience with their bot.

And because of this potential, many innovators and investors think we are soon approaching a "ChatGPT Moment" for robotics where the acceleration and adoption takes off exponentially.

Famed VC investor Vinhod Khosla said recently "By the 2030s, almost everyone will have a humanoid robot at home. Robots won’t be programmed -- they’ll learn tasks by themselves. Humanoid robots will become common because they are cheaper to produce at scale."

Obviously, open-source training "on-premise" (your home) makes it all more realistic. While there are concerns about the "edge device" application safety of a humanoid in your home, we've got a few years to figure that part out, since companies like K-Scale prefer right now to just sell to developers and companies willing to do the work and take the risks.

For more on the robotics revolution, be sure to go here to see all the links, videos, and commentary...

The Week In AI: $9,000 Humanoid Robots, Zuck the Poacher, Larry Page Builds Skynet

Here's a hit list of the other 5 big topics in our weekly run-down...

2) Zuck the Poacher: Buying, Bribing, Stealing Talent from Everywhere

On June 30 Meta Platforms META introduced its new "Superintelligence Lab," led by Alexandr Wang from Scale AI and Nat Friedman. Recall that in late June, Mark Zuckerberg announced a deal to acquire 49% of Scale AI for $14.3 billion. It was seen quickly as the first major “acqui-hire” in the AI race.

The new division includes 11 research hires from top AI labs, including OpenAI, Anthropic, and Google.

Last Friday morning, I also did an interview with Bloomberg about Apple AAPL and the narrative that they are "falling behind" in AI and need to catch-up by proving that Apple Intelligence is real and robust.

I offered the view that Apple is not behind just because Siri isn't performing LLM duties. Tim Cook & Co. will take their time to get that right because they are a "product perfect" company first. And they have their eyes on the prize of being the ultimate healthcare assistant, not an answer bot.

But where Cook could improve his strategy is on recruiting top talent, especially after losing his COO Jeff Williams to retirement and rumors of other infighting that hamper innovation. That's why I told Bloomberg he needs to steal a page from Zuck's playbook.

And a few hours later, we learned of Alphabet's GOOGL magnificent "rearguard" action...

Google Goes Windsurfing in the AI War for Talent

3) "The Great Separation" -- Who Wins, And Who Gets Left Behind, in the AI Age by Coatue Mgmt

Coutue is a cross-asset investment firm specializing in technology with nearly $100 billion AUM. Half is in public Wall Street equities like NVIDIA NVDA and Taiwan Semiconductor TSM and the other half is in Silicon Valley startups from Abacus AI, run by frequent Week In AI expert source Bindu Reddy, to YAHAHA, a metaverse 3D creation platform.

I run through 10 slides from a great presentation they just gave with the guide of Jason Lemkin. Their research is deep and so I am not quick to question their assumptions or conclusions about what is coming with the AI Economy Transformation -- especially since I already agreed with most of their ideas years ago.

This part of the X Space is worth your time alone and comes in around the 25-minute mark. Plus, the comments section of the X Space has the link to Jason's run-through of his top ten slides.

4) JOBS: Never a Shortage of Automation Doomerism

As always for The Week In AI, I collect a handful of the latest studies and expert opinions on the fate of knowledge worker jobs as AI gets better at human numerical and language tasks.

Emad Mostaque of Stability AI says AI agents will outperform humans on most digital tasks by 2026. But AGI won't be a "godmind." 

"It'll feel like your smartest coworker on Zoom or WhatsApp. We don't need polymaths. Just competent workers getting the job done. I want cooks, not chefs." 

In addition to a few other views on the employment outlook, I also look at a post by Amanda Goodall @thejobchick where she breaks down how IBM just replaced 8,000 HR staff by automating 94% of their tasks -- including terminations!

5) Medical Superintelligence

Microsoft introduced MAI-DxO (Diagnostic Orchestrator), what the company is calling a "step towards medical superintelligence."

The system solved 85.5% of 304 cases vs. just 20% by experienced physicians. It also delivered greater cost savings than human doctors.

In the X Space, I share a good 6-post thread with Microsoft graphics by "Chubby" @kimmonismus on X...

Microsoft's LLM is not only designed for multiple-choice questions, but also for real medical diagnoses in realistic scenarios – and outperforms even top models such as o3.

Another great science post last week came from @Dr_Singularity about a “Self-driven Biological Discovery through Automated Hypothesis Generation and Experimental Validation.”

Dr. Singularity said "An autonomous scientific engine has arrived" and I couldn't agree more or be more excited by this! He introduces the research this way:

A new study unveils a fully automated framework that combines:

Large Language Models – to reason, hypothesize, and plan.

Relational Learning – to connect complex biological knowledge.

Robotic Labs – to run experiments in the real world.

All of it working in a closed loop, with zero human input. AI now asks the questions, runs the tests, and learns from the results.

We’re entering a world where machines can do science faster than humans ever could.

Biology is just the beginning.

Welcome to the age of autonomous discovery.

(end of Dr. Singularity's intro)

Gives me chills about what is possible now with the physical sciences!

6) Manhattan Projects for a High-Tech Industrial Renaissance

One big theme I've been tracking lately are the vocal calls by experts for bigger scale industrial projects to support AI and robotics industries. Marc Andreessen and Eric Schmidt are two voices I really respect here.

A few weeks ago I profiled the proposal of Jan Sramek of California Forever to build an "American Shenzhen" high-tech R&D city northeast of the Bay Area. Jan wrote in a post...

"America cannot compete with China without building America's Shenzhen – a place to build drones, ships, robots, and everything else cutting edge. Where? On @CAForever's 100 square miles in Solano, 60 miles north of SF/Silicon Valley, where we invent that stuff."

Then last week I shared a video clip from a great X AI follow @vitrupo quoting General Matter CEO Scott Nolan, who basically says...

"AI demand could match the entire U.S. power grid by 2030. Without urgent expansion of nuclear reactors and fuel, we don't just risk brownouts and skyrocketing energy costs. We risk not getting AI at all."

This is why I just bought Rolls-Royce RYCEY shares as they are getting more contracts to build small modular reactors (SMR) in the UK and Europe.

So when my colleague Ethan Feller brought a paper to our X Space by Epoch AI titled How big could an “AI Manhattan Project” get? I was very excited. 

A man who's wasting no time planning for such projects is Softbank's Masayoshi Son who announced in late June his idea for a $1 trillion "American Shenzhen" in Arizona in conjunction with TSMC.

Ethan also covered the latest project of the other Google founder Larry Page, who has launched a startup called Dynatomics, focused on using AI to revolutionize product manufacturing. 

All this suddenly made me think last week..."AI isn't electricity -- it's gunpowder" in terms of its revolutionary impact on power and wealth.

Be sure to check out The Week In AI X Space for all the links. I love making a curated menu of topics every week where you are guaranteed to find something delicious!

Kevin Cook is a Senior Stock Strategist for Zacks Investment Research where he runs the TAZR Trader portfolio with several AI holdings mentioned here, including NVDA, TSM, AMD and others.

Want the latest recommendations from Zacks Investment Research? Today, you can download 7 Best Stocks for the Next 30 Days. Click to get this free report

This article originally published on Zacks Investment Research (zacks.com).

Zacks Investment Research