Has Claude become a useful tool for bitcoin developers?

[Synchronice]

I only say OpenAI company is against open AI. I don't think OpenAI company is against AI itself.

A.I. has turned into a cut-throat business and the handful of players involved each want to have a monopoly in this industry that is predicted to be a multi-trillion dollar industry. Whoever can gain that kind of power will control infrastructure, labor and even "raw materials" of the future global economy.
Even though China came along with its DeepSeek and destroyed that monopoly but none of them have yet given up competing over that position...

But I think there is a big number of constraints in order for one single company to become the dominant force in A.I. One of those constraints could be geopolitical because I doubt that a Chinese A.I software could ever become the leading application in many countries outside of China. U.S. based citizens, companies or organizations would probably not prefer a Chinese app over other apps even if it is superior, which in my opinion will always prevent one app to ever become largely superior. Perhaps it would be similar to Meta (Insta, Facebook, Whatsapp) vs. WeChat.

I was wondering whether a serious bottleneck in that race to the top could be hardware shortages. It's probably an advantage if an A.I. business is built in a country where chip production is high? Maybe that gives the U.S. an advantage as they have companies like NVIDIA and AMD. Essentially most leading AI chip designers are from the U.S., the largest cloud providers are also from the U.S. and they have export restrictions in place for their most advanced A.I. chips.

China is surely pouring hundreds of billions into A.I. developent and all the hardware manufacturing, but that would still never eliminate the geopolitical aspects for users from around the globe.

It will probably stay a highly competitive sector as the switching costs are low. It is not like a newly launched social network where a user would be dependent on their friends and family to switch as well. If a strong A.I. software comes up, people could just switch and I think that makes it attractive newcomers to try and launch their models whenever they think they are cutting edge.

The US did the right thing by investing into modern technologies. Currently, it has Nvidia and AMD (ironically, origin of CEOs of both companies is Taiwan). Anyways, the USA is great at chip design, EDA software and cloud computing but Netherlands is great at lithography-system design and integration, Taiwan at leading-edge contract manufacturing, Germany at critical lithography optics and so on. So, I would say that the advantage that America has in AI, is really an allied-supply-chain advantage and not something done single-handedly.

What I like about China is that it's trying to compete with open-source projects. Buy some GPUs or RAMs and you can have a great, free AI at home. They won't work as well as Claude but they'll work very well. This can directly hit Claude, ChatGPT, Gemini and other AIs once Chinese open-source AIs advance. Chinese people aren't good at creating something but they are good at copying and making it very cheap/available.

[dzonikg28]

The US did the right thing by investing into modern technologies. Currently, it has Nvidia and AMD (ironically, origin of CEOs of both companies is Taiwan). Anyways, the USA is great at chip design, EDA software and cloud computing but Netherlands is great at lithography-system design and integration, Taiwan at leading-edge contract manufacturing, Germany at critical lithography optics and so on. So, I would say that the advantage that America has in AI, is really an allied-supply-chain advantage and not something done single-handedly.

I think it is well described as allied-supply-chain advantage, but since the USA has the upper hand in so many respects, the final word in so many things, it's really an allied-supply-chain with the USA having a majority vote and a majority share in that chain I guess? I don't know for sure, but given how many parts of this complex industry is dominated by the USA, I think that could be about true. 

What I like about China is that it's trying to compete with open-source projects. Buy some GPUs or RAMs and you can have a great, free AI at home. They won't work as well as Claude but they'll work very well. This can directly hit Claude, ChatGPT, Gemini and other AIs once Chinese open-source AIs advance. Chinese people aren't good at creating something but they are good at copying and making it very cheap/available.

Do you think the underlined part is still true today? No doubt that this was what they did for a very long time and they still may do this here and there, but I think they have their own concepts and smart minds to get things done. China is among the countries with the highest R&D spending in the world. DeepSeek, Alibaba, Tencent, BYD, CATL, all the solar panel business, they have plenty of companies that are world leaders in their branch. But there is one very relevant thing related to this topic that they aren't good at and that is advanced semiconductors and lithography equipment.

Do you think this could influence their strategic considerations regarding an attack on Taiwan? They would still have to solve the lithography equipment shortage as ASML would stop providing machines immediately.

[pooya87]

Do you think the underlined part is still true today? No doubt that this was what they did for a very long time and they still may do this here and there, but I think they have their own concepts and smart minds to get things done. China is among the countries with the highest R&D spending in the world. DeepSeek, Alibaba, Tencent, BYD, CATL, all the solar panel business, they have plenty of companies that are world leaders in their branch. But there is one very relevant thing related to this topic that they aren't good at and that is advanced semiconductors and lithography equipment.

That's right. The "China is copying stuff" is an old view that changed 10-20 years ago. They've invested a ton of money into their technological advancement and have not only caught up but in certain areas they are leading.

In any case I personally wish more countries start working in the AI scene and offer us solutions because at the end of the day even if the industry has 2 giants (US and China aka a duopoly), from our point of view that is pretty much the same a monopoly.

Do you think this could influence their strategic considerations regarding an attack on Taiwan? They would still have to solve the lithography equipment shortage as ASML would stop providing machines immediately.

In my opinion that is all part of their strategy. The Chinese don't consider Taiwan to be another country. They consider it their own soil. If they render certain things (like chip production) that exist in Taiwan useless, then foreign interest in Taiwan (eg. US regime sending them weapons and intervening in their political system to install anti-China people there) would dwindle and separatism would vanish too.

[dzonikg28]

...they aren't good at and that is advanced semiconductors and lithography equipment.

That's right. The "China is copying stuff" is an old view that changed 10-20 years ago. They've invested a ton of money into their technological advancement and have not only caught up but in certain areas they are leading.

In any case I personally wish more countries start working in the AI scene and offer us solutions because at the end of the day even if the industry has 2 giants (US and China aka a duopoly), from our point of view that is pretty much the same a monopoly.

-snip-

In my opinion that is all part of their strategy. The Chinese don't consider Taiwan to be another country. They consider it their own soil. If they render certain things (like chip production) that exist in Taiwan useless, then foreign interest in Taiwan (eg. US regime sending them weapons and intervening in their political system to install anti-China people there) would dwindle and separatism would vanish too.

I think I have to correct myself for that part in my post about China not being able to build lithography equipment. I just stumbled upon a video and then some reports that China did recruit 12 former ASML engineers in a secret operation. That's a crazy move and shows that China will do whatever is necessary to finally break that ASML monopoly on EUV machines.

Former ASML engineers given fake IDs
One veteran Chinese engineer from ASML recruited to the project was surprised to find that his generous signing bonus came with an identification card issued under a false name, according to one of the people, who was familiar with his recruitment.

When it became public that Shanghai Aishengna Electronic ​Technology Group is now close to mass producing DUV lithography machines, ASML's market capitalisation went down by €60 billion. I don't know too much about the technology, but got started going down that rabbit hole and that's really an incredibly complex industry. I assume that China will try to build their own supply chain over time to also be able to build EUV machines, which are even more complex from what I understand. But when they are willing to pay insane wages, they would probably be able to recruit from European companies.

What's a bit confusing are the varying estimations as to when China will be able to compete with ASML. Some say it could be possible in a few years, some say not before late 2030s. I can imagine that it's rater sooner than later because who knows how much is already going on there in secret. China is ambitious and since they know how important this industry is, I believe they may be further down the road than publicly known. They have built things in secret so many times and then surprised the world. But this could indeed be one of their toughest challenges ever as it is said that ASML has a few hundred highly specialised suppliers, many of which are in Europe, have export restrictions and are hard to imitate. Yet, I believe they will get it done, exactly like they did in so many other industries when everybody was shocked when they presented their products out of a sudden.

[BlackHatCoiner]

When it became public that Shanghai Aishengna Electronic ​Technology Group is now close to mass producing DUV lithography machines, ASML's market capitalisation went down by €60 billion.

DUV is mature tech that has been around for decades. Getting domestic DUV into production is a necessary milestone for them, but the gap between DUV and EUV is massive. Spooking the stock market on headlines is easy. Achieving commercial-scale production is another story.

What's a bit confusing are the varying estimations as to when China will be able to compete with ASML. Some say it could be possible in a few years, some say not before late 2030s. I can imagine that it's rater sooner than later because who knows how much is already going on there in secret.

ASML's problem isn't just the internal R&D or having the top engineers; it's the entire ecosystem of highly specialized European suppliers that took 30+ years to refine.

[Danish Ali]

ASML's problem isn't just the internal R&D or having the top engineers; it's the entire ecosystem of highly specialized European suppliers that took 30+ years to refine.

That is right, but this is an ecosystem reasoning, both directions. More important, China does not necessarily have to make a copy of that supply chain, given it cannot get those services at the moment, and can not do that overnight anyway. It may be rough and slow in the beginning but it will (eventually!) be clearer and it will be smoother, 30 years to improve does not mean no one else can start their own 30 years to improve process with full state support pushing them to do so.

[Synchronice]

I think it is well described as allied-supply-chain advantage, but since the USA has the upper hand in so many respects, the final word in so many things, it's really an allied-supply-chain with the USA having a majority vote and a majority share in that chain I guess? I don't know for sure, but given how many parts of this complex industry is dominated by the USA, I think that could be about true. 

I got your point, the US has a very strong influence over countries that I listed above but still, its AI advantage depends on capabilities it doesn't control unilaterally. The US can pressure these countries but it still can't manufacture these things.

Chinese people aren't good at creating something but they are good at copying and making it very cheap/available.

Do you think the underlined part is still true today? No doubt that this was what they did for a very long time and they still may do this here and there, but I think they have their own concepts and smart minds to get things done. China is among the countries with the highest R&D spending in the world. DeepSeek, Alibaba, Tencent, BYD, CATL, all the solar panel business, they have plenty of companies that are world leaders in their branch. But there is one very relevant thing related to this topic that they aren't good at and that is advanced semiconductors and lithography equipment.

Do you think this could influence their strategic considerations regarding an attack on Taiwan? They would still have to solve the lithography equipment shortage as ASML would stop providing machines immediately.

DeepSeek - an AI company. Artificial intelligence as a formal field was largely established in the United States.
Alibaba - Alibaba - online shopping, e-commerce platform. The first kind of one was created in the United States, by Amazon.
Tencent - Services that are into this ecosystem first appeared in the United States. What tencent really did is that it combined many things into one and an example is WeChat.
BYD - Europe and the US worked on electric vehicles back in 1800s first, before China.
CATL is another case where China did not invent the underlying technology first.

I agree with you that China has become far more innovative than it was previously but when it comes to game changer innovations, the US and Europe are still leaders.

[Cinexa]

Attempted to use some of the LLMs on Mac with Ollama, and it was terrifying. It took more than one minute to say just "Hello!".
I thought it is because it was a first time. After an hour working with Ollama, I realized that I need a monster of machine to run LLMs locally.

[Antidote47k]

Attempted to use some of the LLMs on Mac with Ollama, and it was terrifying. It took more than one minute to say just "Hello!".
I thought it is because it was a first time. After an hour working with Ollama, I realized that I need a monster of machine to run LLMs locally.

More than a minute just to answer “Hello” sounds really brutal. Had similar impression of local models until I realize how much your choice of machine and hardware changes the experience.
I am curious, what Mac and model did you use with Ollama? If it is a large model (such as a 32B or 70B) on a base Mac, then it can be expected that Ollama offloaded layers to the CPU or spilled them into swap memory, which completely tanks token generation speed. Lightweight models on almost any silicon Mac should respond almost instantly.

[BobLawblaw]

Generally I’m really interested in hearing how people are integrating it into their workflow.

I'm using Claude Code and Fable 5.1/Opus 5, along with a suite of local LLM's running on a small cluster of DGX Sparks, to help me write an entire AI generated x86 ASM version of bitcoind for Linux that is 100% Core compliant.

[gwestcot]

AI is definitely a useful tool for coders (a lot of other tasks) but it is important for us to first understand what type of a tool it is. The problem is in the name: A.I. which they chose for marketing purposes. But AI is not intelligence. It cannot think. It cannot reason and most importantly it cannot innovate. A.I. is best described as a statistical pattern matcher. But that doesn't sell! The companies couldn't say "come invest millions in our 'statistical pattern matcher'"

After we get that definition correctly, we can move on to utilize this tool more efficiently. Then you wouldn't expect a "pattern matcher" to write you an innovative code solving a new problem. You'd use it to solve a problem that was solved a million times before my many others and that AI has it in its database aka training data.

My question is, for those who are actively building, reviewing code or contributing to Bitcoin-related projects, how are you actually finding it in practice?

I have only used them for code review and have already found one serious bug (a dumb IndexOutOfRange bug) and a bunch of small improvements in my code.

I agree.

I think that just prompting and asking the IA to deliver a full code/app for you won't work in the long run. AI will make mistakes and the app will have a lot of bugs.

But it is useful to make me code faster. I usually ask AI tasks like this: "create a function A that will organize this data into an array and then loop every element using this other function B, and finally organize the response into C". Then I can carefully look at what it did and make just minor adjustments. And this actually saves a lot of time.

I am not a core developer, I just AI to solve problems that have been solved already a million times... But I think a core developer might have some use for it, specially if running locally to protect his privacy and security

Honestly, its best used as almost as a step-by-step assistant rather than having it code massive things all at once. If you keep the scope small and somewhat compartmentalized it seems really good but if you give it a massive scope, then even now it seems insufficient. It is also only as good as the prompts you give it.

[alegotardo]

Honestly, its best used as almost as a step-by-step assistant rather than having it code massive things all at once. If you keep the scope small and somewhat compartmentalized it seems really good but if you give it a massive scope, then even now it seems insufficient. It is also only as good as the prompts you give it.

I have been using AI in my day-to-day work and I completely agree with what you said... I ca not blindly trust to AI a complex work and believe that it did everything correctly.
I usually ask AI to help me with automated test cases or to do a "repetitive work" and even then I need to keep monitoring everything it write, because if I tell to it: Do not do "X", do "Y"... it will do "Y" as I ordered, but after some time or if it believes it has found a big problem, it will try to find a solution and sometimes goes back to doing "X". So, it does not always depend only on the efficiency of my instructions, it seems to have a "life of its own" sometimes and that is not a good thing.

[gwestcot]

Honestly, its best used as almost as a step-by-step assistant rather than having it code massive things all at once. If you keep the scope small and somewhat compartmentalized it seems really good but if you give it a massive scope, then even now it seems insufficient. It is also only as good as the prompts you give it.

I have been using AI in my day-to-day work and I completely agree with what you said... I ca not blindly trust to AI a complex work and believe that it did everything correctly.
I usually ask AI to help me with automated test cases or to do a "repetitive work" and even then I need to keep monitoring everything it write, because if I tell to it: Do not do "X", do "Y"... it will do "Y" as I ordered, but after some time or if it believes it has found a big problem, it will try to find a solution and sometimes goes back to doing "X". So, it does not always depend only on the efficiency of my instructions, it seems to have a "life of its own" sometimes and that is not a good thing.

It depends on the model you are using, but long-term memory is key. I think it has actually improved a ton over the last year or so in this regard. Prompt discipline is also important so that it can stay focused. If I want to discuss something secondary or tertiary, then it ends up being in its own chat rather than the main chat.