In this episode, Sunny Ray reconnects with longtime Bitcoin developer Peter Todd for a wide ranging conversation that begins with artificial intelligence and ends with the accidental origin of some of Bitcoin's most familiar terminology. Todd argues that modern AI is essentially sophisticated pattern matching without real logical reasoning, the opposite of the sci fi vision of thinking machines, and explains why he believes true human level AI remains an open and possibly distant question. The conversation shifts into Todd's eclectic background, from a fine arts degree and a dropped physics program to years spent doing analog electronics design, including building ultra sensitive gravity sensors using superconducting squid controllers. Todd then traces his path into Bitcoin, which he first learned about in 2009 but did not take seriously until 2013. He shares the surprisingly casual stories behind terms like proof of sacrifice, Merkle mountain ranges, and coinjoin, terms he either coined or helped popularize during Bitcoin's early, freewheeling era. Throughout, Todd offers a grounded, skeptical engineer's perspective on hype, whether about AI or crypto all time highs.
Bitcoin OG Peter Todd breaks down AI's real limits and the accidental origins of Bitcoin's most famous terms.
What are your thoughts on AI?
I am kind of disappointed at where we are. All our AI stuff is the total opposite of what we expected. It is about pattern matching and it is actually kind of dumb in some ways even though it seemingly performs miracles. It is the opposite of the Star Trek Data AI that has logical reasoning power. AI is exactly opposite, it has pseudo fake human touch and no logic.
Do you think we are close to general AI coming together?
I read Kai Fu's book and he says you need at least five or six big breakthroughs to actually make this happen, and we are only close to maybe one. Part of the breakthrough is understanding what intelligence even is. We have no idea how well these pattern matching systems map to how the human brain works, so I would not say we are far away, I would say we have no idea how far away we are.
Could AI actually cause something like nuclear warfare?
Nuclear warfare in the hands of AI probably cannot happen in a simple way because these systems are not likely to actually be connected to the networks that matter. Look at the old Russian doomsday mechanism, it was not automated, it gave low level humans in bunkers firing codes they normally would not have. The way AI would actually pull off nuclear war is not by hacking systems, it is by fooling human beings.
Has anyone tried to impersonate you using AI?
Someone just tried to impersonate me on Telegram. I had someone send me the text and it sounded nothing like me at all, it obviously was not written by AI, it was written by a human who did not know what they were doing. Ironically, an AI could have done that job better if it were fed everything I have written online, it would have sounded more convincing than what the scammer put together.
What is your Bitcoin origin story?
I honestly do not quite remember how, but I somehow learned about Bitcoin in early 2009 while doing a physics degree and working full time, so I did not pay much attention, which was the greatest financial mistake of my life. I became more active around 2013 and started looking at Bitcoin stuff more seriously from there.
How did you end up coining some of Bitcoin's terminology?
There are multiple terms in Bitcoin where I am the guy who came up with the name, like proof of sacrifice, and Merkle mountain ranges, where I was yet another person to rediscover the concept but my explanation and name stuck. Coinjoin is another example, Greg Maxwell actually asked me to come up with a good name for it.
Can you explain the gravity sensor work you did in analog electronics?
We were building a gravity sensor for an airplane, and since there is no zero force reference frame in turbulence, we measured the gradient of the gravitational field between two test masses instead of gravity directly, canceling out turbulence forces. The displacement we needed to measure was about a hundredth the width of a hydrogen nucleus, so we used a superconducting squid controller since magnetic fields were the only way to detect movement that small.
Why did you switch from industrial design to fine arts?
I switched after my teachers turned out to not understand basic physics. One presented a gravity powered lamp as an award winning environmental design idea, and a quick calculation showed a twenty pound weight falling six feet could not power a lamp meaningfully. Realizing the teachers were clueless about actual engineering, I figured I might as well just go into fine arts.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview