Are Humans a Stepping Stone in Evolution for Robots?
Andrew Corp
Introduction
As artificial intelligence and robotics advance, we are forced to confront profound questions about the future of humanity and our role in the evolution of intelligence. This post explores a thought-provoking question posed to Sergey Brin: Are humans simply a stepping stone in evolution for robots and silicon intelligence?
I was watching a panel where Sergey Brin was asked by David Friedberg basically "Are humans a stepping stone in evolution for Robots (Silicon Intelligence)?" https://youtu.be/8g7a0IWKDRE?t=1300
It really made me take a step back and think about it. Robots will be able to do everything faster than humans, there is no question about that. The jury is still out if it'll be better. Robots will likely also not have emotions (it's ultimately up to the creators of the robots if they do or don't). Given, emotion slows down performance.
I took some time to think about why robots will be better and thought about Communication and how Robots will communicate to eachother. It's very different than humans. Humans need to open their mouths and let out sounds. Whereas robots will really use connections or likely some sort of Websocket-like technology. Digging deeper into that, humans can't really listen to two conversations at one time. Computers can handle MANY connections at one time https://grok.com/share/bGVnYWN5_a8f9632e-75c0-4fb8-a19d-0898c4e7d032 Lets take the really really low number here of 1000. This means that robots can talk and listen to 1000 different people in parallel (1000x times what humans can do). Can you imagine talking to 1000 people on the phone at once where everyone is talking?
What would robot-to-robot communication look like?
I landed on some key-requirements:
- Secure Connections (Encryption, Authentication, Integrity)
- The conversation doesn't ALWAYS need to be secure, a person can stand in a park and yell stuff.
- Private conversations need to be secure.
- Bi-directional data transfer
- I want to speak and listen
- The messages could be in many different formats
- These are basically the language you are speaking (Spanish, Italian, English, ...)
- Non-dependent on Internet
- If you are talking to someone next to you, does it make sense to depend on the internet tell the person literally 1 foot away from you something? I think it could, if for security purposes you want to log all conversations.
- Decentralized
Key Challenges
- Scalability: Supporting thousands of simultaneous connections in a decentralized mesh.
- Security: Ensuring privacy, authentication, and integrity in both public and private robot conversations.
- Interoperability: Handling multiple data formats and communication protocols between heterogeneous robots.
- Resilience: Maintaining robust communication even as nodes join/leave or move in/out of range.
- Low-latency: Achieving real-time or near-real-time data transfer for mission-critical tasks.
It's a very interesting problem to solve. it's like having a Mobile LAN where Nodes can connect and disconnect quickly. The obvious technology that comes to mind is Bluetooth it doesn't fit all of the above, definitely not 1000 simultaneous connections.
I think using Zigbee and LoRa (not the LLM term, but the Long Range Wireless Technology) it's possible to make this happen.
Link to the simulation: https://gemini.google.com/share/21076fac294d
Rust would likely be the language of choice to implement this in. Hardware interaction and using Zenoh for the transfer/application layer.
Because the payload is agnostic, you could use JWTs if you wanted to, to protect agains man-in-the-middle attacks.
Potential Impact
I am thinking about trying to build a prototype. There are broad industries where this could be useful, just to name a few:
- Smart Street lights (Cars talking to street lights)
- Cars talking to each-other
- Robots in warehouses communicating with each-other
- Humanoid's when they are trying to accomplish tasks that are dependent on each-other