First eyeballs, now brains. Sam Altman is hungry | 3 Wow and 1 Promise

Summary

This episode of Turing Post explores three major developments in AI and one overarching promise about the future of human-machine integration.

Highlights

”This might be the last decade when your thinking is not recorded”

yt-dlp --download-sections "*0:00-0:55" "https://www.youtube.com/watch?v=bXD4-e9FUIE" --force-keyframes-at-cuts --merge-output-format mp4 -o "bXD4-e9FUIE-0m00s.mp4"

“First eyeballs and now brains. Sam Altman is really hungry and now rivals Elon Musk in the brain department. Your brain literally. You think AI can surprise you? Think again. And remember, this might be the last decade when your thinking is yet not recorded and analyzed.” — Turing Post, 0:00

”Sam Altman’s Cyber-Physical Ecosystem”

yt-dlp --download-sections "*1:32-2:30" "https://www.youtube.com/watch?v=bXD4-e9FUIE" --force-keyframes-at-cuts --merge-output-format mp4 -o "bXD4-e9FUIE-1m32s.mp4"

“The eyes, the identity, the brain, the devices you wear like the second skin, the data streams feeding robots, every piece drawn together into one vast connected system. Everything that can be merged. Sam Altman aims to merge on a planetary scale shaping an all-encompassing softly guided singularity.” — Turing Post, 1:32

”The merge is probably our best case scenario” - Sam Altman, 2017

yt-dlp --download-sections "*3:20-4:20" "https://www.youtube.com/watch?v=bXD4-e9FUIE" --force-keyframes-at-cuts --merge-output-format mp4 -o "bXD4-e9FUIE-3m20s.mp4"

“The merge can take a lot of forms. We could plug electrodes into our brains or we could also just become really close friends with a chatbot. But I think a merge is probably our best case scenario.” — Sam Altman (2017 blog post), 3:47

”The language of our mind could end up in the hands of one corporation”

yt-dlp --download-sections "*4:49-5:45" "https://www.youtube.com/watch?v=bXD4-e9FUIE" --force-keyframes-at-cuts --merge-output-format mp4 -o "bXD4-e9FUIE-4m49s.mp4"

“The most secret thing we have, the language of our mind could end up in the hands of one person or one corporation. And the question is as old as the earth itself. Can we nurture such a powerful technology, such a powerful creation, and still remain the masters of our own destiny?” — Turing Post, 5:05

”Three things you can watch forever: Fire, water, and a robot folding laundry”

yt-dlp --download-sections "*5:37-6:40" "https://www.youtube.com/watch?v=bXD4-e9FUIE" --force-keyframes-at-cuts --merge-output-format mp4 -o "bXD4-e9FUIE-5m37s.mp4"

“In the world, there are truly three things you can watch forever. Fire burning, water falling, and a robot folding your laundry… And while it works, it even meets your gaze, gestures naturally, as if it were quietly saying, See, I can learn this too.” — Turing Post, 5:50 The central focus is on Sam Altman’s expanding technological empire, which now includes Merge Labs - a brain-computer interface company funded by Tools for Humanity (the same organization behind Worldcoin’s iris-scanning technology). The video draws parallels between Altman’s various ventures, suggesting he is building a comprehensive “cyber-physical ecosystem” that encompasses human identity verification (through eye scans), brain-machine interfaces, and AI systems.

The episode contrasts Merge Labs with Elon Musk’s Neuralink, noting that while Neuralink focuses on restoring capabilities for people with paralysis, Merge Labs appears aimed at broader human-machine merging for healthy users. The video highlights Audrey Cruz, a Neuralink patient who has been sharing her raw, unfiltered progress on social media. A key technological development discussed is “brain foundation models” - AI systems trained on brain scan data (like BrainOmni) that could eventually decode neural signals faster and more accurately, especially when trained on invasive neural recordings.

The video also covers two additional “wows”: Figure 01’s humanoid robot Helix learning to fold laundry through experience, and Skywork AI’s Matrix Game 2.0 - an open-source, real-time world model that generates interactive environments at 25 frames per second. Together, these developments point toward a future where brain-machine interfaces, embodied AI, and interactive world models converge into a new human-AI ecosystem.

Key Points

• Brain Computer Interface Race (0:23) - The BCI field is heating up with two major competitors: Neuralink led by Elon Musk and Merge Labs backed by Sam Altman
• Merge Labs Funding (0:53) - Merge Labs is funded by Tools for Humanity, the same company behind Worldcoin’s iris-scanning digital ID project
• Sam Altman’s Vision (1:14) - Altman thinks in terms of planetary-scale systems and builds technology at the intersection of human identity and data
• Cyber-Physical Ecosystem (1:32) - Eyes, identity, brain, wearable devices, and robot data streams all connecting into one vast system that Altman aims to merge on a planetary scale
• Neuralink vs Merge Labs Goals (2:25) - Neuralink focuses on restoring capabilities for paralysis patients while Merge Labs aims for broader human-machine merging
• Audrey Cruz’s Journey (2:44) - A Neuralink patient sharing raw, unfiltered progress including cursor control through thought and writing her name mentally
• Altman’s 2017 Prediction (3:35) - Sam Altman wrote about the merge between humans and machines on his blog in 2017, connecting it to his work at OpenAI
• Brain Foundation Models (4:03) - New AI models like BrainOmni are being trained on EEG and MEG data to learn general patterns in brain activity
• Invasive Neural Recordings (4:32) - Training brain foundation models on invasive recordings from Neuralink and Merge Labs could enable faster, more accurate brain signal decoding
• Privacy Risks (5:08) - The language of our minds could end up controlled by one person or corporation, raising fundamental questions about human autonomy
• Helix Robot Folds Laundry (5:42) - Figure 01’s humanoid robot learned to fold towels through experience alone, with no changes to its programming
• Three Things to Watch Forever (5:50) - Fire burning, water falling, and a robot folding your laundry
• Matrix Game 2.0 (6:38) - Skywork AI’s open-source real-time streaming world model generates fully interactive environments at 25 FPS
• Three Key Technical Moves (7:37) - Matrix Game 2.0 uses 1,200 hours of training data, action injection for direct input, and causal autoregressive diffusion for efficiency
• Multi-Environment Support (8:14) - The system works across Minecraft, GTA 5, Tempo Run, and custom environments with smooth motion and responsive scenes
• Beyond Gaming Applications (9:01) - World models can simulate environments for robotics, AR/VR, and medical assistance
• Open Source Availability (9:12) - Matrix Game 2.0 is open source and available for anyone to use immediately
• Theory to Practice Shift (9:19) - Interactive AI worlds are no longer lab demos - they work and are ready for exploration

Mentions

Companies

• Merge Labs (0:30) - New BCI startup backed by Sam Altman aiming for broader human-machine merging
• Neuralink (0:27) - Elon Musk’s brain-computer interface company focused on restoring capabilities for paralysis patients
• OpenAI (0:35) - AI company co-founded by Sam Altman, connected to his vision of human-machine merging
• Worldcoin (0:35) - Crypto project that scans eyeballs in exchange for digital ID, co-founded by Sam Altman
• Tools for Humanity (0:58) - Company that funds Merge Labs and built Worldcoin
• Figure 01 (5:45) - Robotics company that created the Helix humanoid robot
• Skywork AI (6:48) - Creator of Matrix Game 2.0 open-source world model

Products & Technologies

• BrainOmni (4:10) - Brain foundation model trained on EEG and MEG data for decoding mental states
• Brain Foundation Models (0:40) - AI models trained on brain data, described as “GPT for brain data”
• EEG and MEG (4:15) - Non-invasive brain scanning technologies used to train brain foundation models
• Helix (5:42) - Figure 01’s humanoid robot that learned to fold laundry through experience
• Matrix Game 2.0 (6:45) - Open-source real-time streaming world model from Skywork AI
• GTA 5 (7:41) - Video game used for training Matrix Game 2.0 with 1,200 hours of interactive video
• Unreal Engine (7:43) - Game engine used alongside GTA 5 for training data
• Minecraft (8:17) - One of the game environments supported by Matrix Game 2.0
• Causal Autoregressive Diffusion (7:58) - Technical process used by Matrix Game 2.0 for efficient real-time generation

People

• Sam Altman (0:00) - CEO of OpenAI and co-founder of Worldcoin, backing Merge Labs for brain-computer interfaces
• Elon Musk (0:25) - Founder of Neuralink, competing in the BCI space
• Audrey Cruz (2:44) - Neuralink patient sharing her journey on social media, demonstrating cursor control and writing through thought
• Paul and Katie (8:37) - Friends of the presenter working on a video game who could benefit from Matrix Game 2.0

Surprising Quotes

“This might be the last decade when your thinking is yet not recorded and analyzed.” — 0:13

“The merge can take a lot of forms. We could plug electrodes into our brains or we could also just become really close friends with a chatbot. But I think a merge is probably our best case scenario.” — 3:47 (Sam Altman, 2017 blog post)

“The most secret thing we have, the language of our mind could end up in the hands of one person or one corporation.” — 5:10

“In the world, there are truly three things you can watch forever. Fire burning, water falling, and a robot folding your laundry.” — 5:50

“Can we nurture such a powerful technology, such a powerful creation, and still remain the masters of our own destiny?” — 5:21

Transcript

0:00 First eyeballs and now brains. Sam Altman is really hungry and now rivals Elon Musk in the brain department. Your brain literally. You think AI can surprise you? Think again. And remember, this might be the last decade when your thinking is yet not recorded and analyzed.

0:17 It’s deeply fascinating that this is happening now in the world of brain computer interfaces, BCI. The race is heating up. And in one lane, Elon Musk and Neuralink. In the other lane, the newcomer, the Merge Labs, backed by the veteran Sam Altman, who started OpenAI and Worldcoin. And looming behind it is a fresh AI twist that could change the game entirely. Brain Foundation models. We see two in one here. A big wow contains a big promise. Let’s go.

0:53 So, let’s start with Merge Labs. This is not just another BCI startup. It’s funded by Tools for Humanity. If you remember, that’s the company that Sam Altman co-founded to build Worldcoin, a crypto project that scans your eyeballs in exchange for a digital ID. All eyeballs on Sam Altman folks. That alone tells you at least two things. First, here is a founder who thinks in terms of planet scale systems. Second, he’s not shy about building technology that sits in their intersection of human identity and data.

1:32 And remember the episode about the cyber physical ecosystem where we traced Sam Altman’s see-through plans? Now it’s right in front of us. The eyes, the identity, the brain, the devices you wear like the second skin, the data streams feeding robots, every piece drawn together into one vast connected system. Everything that can be merged. Sam Altman aims to merge on a planetary scale shaping an all-encompassing softly guided singularity, a living cyber physical ecosystem.

2:04 It’s really something huge. It’s a vision that feels 10 steps ahead of almost everyone in the field. Not just building a product but arranging the building blocks of an entire future. One where the merge between humans and technology is not a distant idea but an operating principle. And I mean Merge Labs - it’s quite a telling name.

2:25 How do they compete with Neuralink? First of all, they both build brain computer interface devices, but if Neuralink’s vision is to restore capabilities, think like helping people with paralysis move again, Merge Labs seems to be aiming for broader human machine merge. As we said this week we saw a profound example of this from Audrey Cruz, the patient with Neuralink’s brain computer interface.

2:49 On her social media that she just started, she began sharing her journey. No fancy press releases, no curated demos, just raw unfiltered progress. We see videos of her through pure thought, working on her accuracy and speed in moving a cursor. We see a photo of her attempting to write her own name for the first time with just her thought. She shares pictures of the small healed implant site captioned before unwrapping and after. It’s an intimate moment by moment look at what the technology actually means for a person.

3:20 This is about high bandwidth connections between the brain and AI systems potentially giving healthy users entirely new forms of interaction. This has clearly been an interest of Altman’s for a while. He wrote about the merge between humans and machines on his personal blog back in 2017 and connects it pretty directly to his work at OpenAI.

3:47 Here’s a quote: “The merge can take a lot of forms. We could plug electrodes into our brains or we could also just become really close friends with a chatbot. But I think a merge is probably our best case scenario.” Of course you do.

4:01 Here’s where it gets really interesting. Researchers have been developing what they call brain foundation models. Think of them like GPT for brain data. One of the latest examples called BrainOmni, and it was trained on mountains of data from EEG and MEG. It learned general patterns in brain activity and can adapt to specific tasks quickly whether it’s decoding mental states or detecting neurological disorders.

4:28 But these models are built from non-invasive brain scans. Imagine training them on the invasive neural recordings - the kind of things that Neuralink and Merge Labs will collect. Suddenly your AI can decode brain signals faster, more accurately and across different people without starting from scratch. This is huge.

4:49 And what a marvel can that be. Just imagine our thoughts flowing faster. Our connection with machines as natural as speech. New abilities opening before us. Devices becoming smarter, cheaper, more affordable, more tuned in with us. It’s a dream from the pages of science fiction coming true.

5:05 Of course, there is a risk. It’s always in nature. Actually, the most secret thing we have, the language of our mind could end up in the hands of one person or one corporation. And the question is as old as the earth itself. Can we nurture such a powerful technology, such a powerful creation, and still remain the masters of our own destiny? There’s a lot of questions like this in AI. We’ll need answers, but the possibilities, these merging…

5:37 All right, here’s your second wow. Much more meditative. Helix from Figure 01 learned how to fold laundry. Let’s watch it. In the world, there are truly three things you can watch forever. Fire burning, water falling, and a robot folding your laundry.

5:58 Folding laundry for us, a simple homely chore that you hope every day you hope your 8 year old will be able to do - in vain. But for a humanoid robot, it’s even more complicated. It’s like taming a restless creature. The cloth twists, slips away, folds in an unexpected way. And yet here it is, a machine with many fingers, folding a towel on its own for the very first time. No change to its mind at all. Only new experience, new memories.

6:25 And while it works, it even meets your gaze, gestures naturally, as if it were quietly saying, “See, I can learn this too. Maybe even better than your 8 year old child.”

6:38 This one is for anyone who ever dreamed of AI that doesn’t just watch the world but lives in it. Matrix Game 2.0 from Skywork AI is one of those moments where you feel the ground shifts under the whole field. It’s an open-source real-time streaming world model. An AI that can generate a fully interactive environment as you move, think, and act.

7:05 No pre-rendering, no waiting. At 25 frames per second, it responds to the instant move of your mouse or a stroke of a keyboard. This is just incredible. Until now, AI world models, the ones that built to simulate environment, were sluggishly slow. You’d make a move, then wait for the system to catch up. The lag killed any sense of presence. And that’s what Matrix Game 2.0 tries to change.

7:32 They pull it off with three key moves. First, scale of experience - over 1,200 hours of interactive video from GTA 5. I love GTA and Unreal Engine, giving the model a vast range of examples to learn from. Second, direct connection - an action injection model that feeds your inputs straight into the frame by frame generation. How cool is that? And three, efficiency in motion. The causal autoregressive diffusion process that predicts what happens next in just a few steps. Instead of the slow grind of hundreds.

8:11 The result is a system that is not tied to one world or one loop. You can switch between Minecraft, GTA 5, Tempo Run, or entirely custom environments. Motion stays smooth. Objects hold their shape. Scenes respond fluidly. You can alter what’s in front of you second by second and watch the changes take effect immediately.

8:31 I’m thinking about my friends who are working on an impressive video game. Hi Paul. Hi Katie. This type of tool is just like a gift for them. It opens up a completely new imagination world of creation in the video games. And I’m thinking about my sons who constantly talking to me about video games. But this is a completely different situation when you’re not stuck in one game or in one world, but everything changes as you move.

8:58 And this isn’t just for play or for games. These world models can simulate environments for robotics, for AR and VR, for emulating and simulating worlds, for medical assistance, etc. And because it’s open source, anyone can use it starting now. That’s very, very, very impressive.

9:19 In my opinion, Matrix Game 2.0 marks a shift from theory to practice. And interactive AI worlds are no longer a lab demo or a teaser. They’re here, they work, and they’re ready for anyone to explore. It’s super exciting.

9:32 Thank you for being with me during this three WOWs and one promise. As always, every week in AI is just amazingly filled with new inventions, with new ideas, with new capabilities, and this frontier is just opening up in front of us, and we are the pioneers. This is super exciting. Subscribe to the channel and leave your feedback. Thank you.