Bitcoin and AI

I’m very interested in Bitcoin and as a software developer am interested in AI (it’s going to affect software development, companies and life in general and I have a lot to learn about it). I was happy to see people talking about the intersection of them.

I wanted to share some links I’ve come across. I don’t have any new insights, but will continue to add to this as I hear more.

Bitcoin – The Natural Choice For Artificial Intelligence

Swan Bitcoin Newsletter Vol. 34

“AI requires a significant amount of computing power today. Asking ChatGPT a single question is estimated to incur 36 cents in computing costs.

Of course, our AI assistants will do much more than just answer our questions. They’ll also be handling many of our annoying chores. To be effective, they will need to pay for storage, computing resources, internet bandwidth, and more.

Not to mention your personal AI assistant will likely need to hire specialist AI agents to perform niche tasks, such as creating a new website for your side hustle.

Consequently, AI Agents will require money to fulfill our requests.”

“Bitcoin is the perfect money for AI agents

There are at least three requirements for money used by an AI agent.

Permissionless - AI agents aren’t people and don’t have identities Instant settlement - AI agents need to instantly allocate resources without the risk of chargeback Micropayments - AI agents need to send tiny fractions of a cent in real time (streaming money)”

Use my Swan Bitcoin referral code if you’re interested in joining Bitcoin Swan

Lyn Alden - The Intersection of AI and Bitcoin

I’ve listened to Lyn Alden on What Bitcoin Did and am hoping to read her book Broken Money in the future. Lyn Alden - August 2023 Newsletter: Six AI Themes to Consider has Theme 6) The Intersection of AI and Bitcoin is long and in-depth. Here are a few highlights.

“Which money is the best form of money for AI agents to use? From what I’ve seen so far, I think bitcoin is the leading answer.”

“The answer is that Bitcoin/Lightning is uniquely well-suited to working with AI”

“Although we’re not quite at the stage yet, in a world of sufficiently advanced artificial intelligence and virtually costless impersonation or forgeries, we might learn to demand proof-of-continuity (cryptographic public/private key pairs) and proof-of-work (evidence that real-world resources were expended to add weight to a given public/private key pair or online identity), before taking certain types of digital content seriously.”

Podcast

BTC148: Bitcoin and AI with Guy Swan - 20 September 2023 Here are some notes I took, it’s a wide ranging conversation so I recommend listening to it on Fountain and earning sats while you do so.

• AI is a centralizing force

“made me realize that there’s going to be a major problem with the way that our payment systems and our monetary structure works, like just a credit based system works and where we are in kind of our digital transition and what AI models and both image generation, video generation and audio generation is going to really mess with all of the tools we have For proving that you are human online, I think we essentially lose that ability, and I don’t think the timeline is very long on that.”

• the fraud possibility is huge

LLMs might become the interface, more like the keyboard and the mouse than Photoshop

LLMs need good data, not just more data. LLMs can run on a laptop, “good enough” but not as good as GPT. You download the “weights” of the trained model, similar to a compression algorithm where it stores relationships between words and phrases.

Things are going to change.

Need encryption models?

OpenAI is running out of money right now, unless they can find a new way to make money

If we are moving to more specialized LLMs over a large generalized model, then payments between each other will really be needed. -> need for Bitcoin!

KYC (Know your customer) and

fraud issues that merchant services need to deal with doesn’t exist with Bitcoin payments on the Lightning Network. 30 day delay for settlement in current system vs seconds.

LORA - Low Rank Adjustments - all people with less powerful hardware can start training micro-trainings, then you can put them all together and make better larger models. A group is better than one institution LLMs will make us more agile and iterate faster. Large companies/models will have trouble keeping up. Focus on putting the pieces together.

he mentioned: https://www.semianalysis.com/p/google-we-have-no-moat-and-neither - out of date already.

Hugging Face “a GitHub of models”

at 1:27:30 “have us think about the way we are thinking”, what is the premise of why I am thinking, threads of thoughts, math of probabilistic connections, abstract a new layer