<\/p>\n
\u201cIt\u2019s good at answering Bitcoin- and economics-related questions \u2014 at least better than GPT-4,\u201d Aleksandar Svetski told Cointelegraph at a bustling Bitcoin Amsterdam.<\/p>\n
The entrepreneur, author and founder of Spirit of Satoshi, a novel artificial intelligence (AI) large language model (LLM), begins to unpack the arduous journey his small startup has undertaken to create its Bitcoin-centric AI chatbot.<\/p>\n
The model is the result of a time-consuming training process to generate responses based on reputable Bitcoin resources, the\u00a0Austrian school of economics<\/a> and libertarian ideals. Still in its infancy, Spirit of Satoshi reflects ideals from a \u201cwell-curated Bitcoin corpus,\u201d including resources such as Saifedean Ammous\u2019 best-seller The Bitcoin Standard.<\/em><\/p>\n Svetski explained that the major difficulty in building the model was not just curating relevant sources of information from books, research papers and podcasts but also guiding the model to generate responses through an exhaustive training process. He added that a common misconception about LLMs is that they\u2019re sourcing information like a search engine:<\/p>\n \u201cThey are just probabilistically stringing words together in a way that is representative of the patterns within the model. So, it\u2019s not even sourcing anything.\u201d<\/p><\/blockquote>\n It\u2019s part of the reason why AI chatbots tend to \u201challucinate\u201d from time to time, Svetski explained, and why developing an LLM requires a focus on training it on a style of answering. Spirit of Satoshi is by no means perfect either, at least not in its current iteration.<\/p>\n \u201cOur model will also hallucinate. It\u2019s also going to talk shit, but it\u2019s going to say something more like a Bitcoiner would say.\u201d<\/p><\/blockquote>\n Having established a broad but targeted base of Bitcoin-centric information and data, Svetski\u2019s team has set about feeding the model tens of thousands of question-and-answer pairs using programmatic methods. However, a human element is still required to help Spirit of Satoshi generate responses that might have come from its namesake.<\/p>\n Related:<\/em><\/strong> Bitcoin Amsterdam: BTC shines in depths of crypto bear market<\/em><\/strong><\/a><\/p>\n The ongoing development of the model is leaning on the wider Bitcoin community as a result. Spirit of Satoshi employs an incentive process that allows the public to verify, create and validate data for the model.<\/p>\n Using credentials from the Lightning Network, Nostr or email addresses, a \u201cproof-of-knowledge\u201d mechanism allows users to get paid in satoshis<\/a> for helping train the model.<\/p>\n The process uses a consensus model that will automatically impose a penalty if users are creating \u201cjunk data.\u201d Svetski described it as the crucial \u201chuman\u201d element to improve Spirit of Satoshi\u2019s outputs:<\/p>\n \u201cIt\u2019s producing incredible content; it\u2019s the last piece to take your content from 80% good to 95% good. And that has a huge impact on the quality of the model.\u201d<\/p><\/blockquote>\n