Humanoid robots get closer to real jobs

Atlas steps onto the factory floor

A humanoid robot sorted car parts under factory lights

Boston Dynamics is testing its Atlas robot at Hyundai’s Georgia plant, training it to handle roof racks on an assembly line using AI and simulation rather than hand-written code. The current Atlas is 5-foot-9, weighs 200 pounds, runs on Nvidia chips, and was trained alongside more than 4,000 digital copies inside a simulator for six-hour runs.

What stands out is how ordinary the work is, not the acrobatics, because sorting parts at scale is where robots stop being demos and start becoming labor. This feels less like a moonshot and more like a slow, expensive grind toward reliability, which is exactly how factory tech actually ships.

Goldman Sachs pegs the humanoid market at $38 billion within a decade, and Hyundai owning 88% of Boston Dynamics explains why this is happening inside a real plant, not a lab. China’s state-backed push adds pressure to move faster, even if the robots still need heavy supervision.

When these robots show up every shift, who decides what work is left for people?

These earbuds are designed for noisy rooms

Voice AI startup Subtle just launched $199 wireless earbuds that use its noise isolation models to make calls, dictation, and transcriptions work in loud places. The buds ship with a one-year iOS and Mac subscription and use a custom chip to wake a locked iPhone, with Subtle claiming five times fewer transcription errors than AirPods Pro 3 paired with OpenAI’s model.

The interesting move here is bundling software ambition into hardware, because Subtle is really selling permission to use voice around other people. If whispering into earbuds actually works, that removes one of the biggest social blockers to voice as an interface.

This puts pressure on dictation apps like Wispr Flow and Superwhisper, which live entirely in software and depend on existing mics. It also hints at a future where AI interfaces fragment into dedicated devices instead of staying trapped inside phones.

If voice finally works in public, do people actually want to talk to computers all day?

Samsung ties its AI future to Google

Samsung wants Gemini everywhere.

Samsung says it will ship 800 million mobile devices running Google’s Gemini AI in 2026, up from about 400 million last year, according to comments from co-CEO TM Roh in Seoul. That includes phones and tablets, rolled out while engineers deal with a memory chip shortage and executives walk past glass offices lit late at night.

This is less about flashy features and more about distribution, because putting the same AI stack across that many devices quietly hands Google a massive consumer footprint. Samsung looks willing to lean hard on Gemini to regain ground from Apple, even if it means ceding some control over the AI layer.

The move strengthens Google’s position against OpenAI just as Gemini 3 and GPT-5.2 are trading blows, and it pressures rivals to respond at similar scale. It also raises the stakes if chip shortages force Samsung to raise prices on hardware carrying those features.

If Gemini becomes default on hundreds of millions of phones, how easy is it to switch later?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.