AI: A Look into the Future with Marcus Graichen
Artificial intelligence has been a hot topic in the tech world for some time now, but what does the future hold? We sat down with Marcus Graichen, a developer and product architect, to discuss his unique perspective on the space. Graichen is a leading educator on Bittensor, a decentralized AI network that incentivizes participants to train and operate over 4,000 machine learning models in a distributed manner.
Decentralized and Open Source AI Tools
Graichen believes that decentralized and open source AI tools, like Bittensor, could become serious competitors to commercial large language models such as ChatGPT. Bittensor operates similarly to Bitcoin, with a transferrable and censorship-resistant token called TAO. It runs on a 24/7 decentralized blockchain substrate, providing transparency and auditability. Miners, like those in the Bitcoin network, can exist globally and anonymously.
Understanding Bittensor
Bittensor may sound complicated, but Graichen compares it to BitTorrent and downloads. Essentially, it is a multi-layered protocol that allows for decentralization and incentivization through a blockchain. The Opentensor Foundation, a non-profit, created Bittensor, and it is now controlled by the foundation. However, the goal is for it to be an open-source technology.
Bittensor Paradigm
Graichen's involvement with Bittensor began when he became interested in cryptocurrency in 2017. He saw its potential to disrupt industries he had been involved in and became a leading educator on Bittensor. He founded taostats, a popular website for Bittensor metagraph statistics and data analysis. He then went on to found BitAPAI, now known as Corcel, which aims to provide developers with access to various Bittensor sub-nets and build their own sub-nets.
How Bittensor Differs from Commercial Language Models
Bittensor is not just a language model; it has two sub-nets that produce text responses. One allows miners to run any model they can get their hands on, including commercially available ones like ChatGPT. Miners are rewarded for diverse responses, while the other sub-net focuses on speed and accuracy. Graichen uses the example of BitTorrent to explain how Bittensor rewards miners for performance.
Digging Deeper into Generative AI
Graichen also explains how Bittensor's genAI products differ from commercial language models. Instead of relying on one large language model, users have access to many models, and the best performers are rewarded. However, he notes that judging the quality of responses from language models can be difficult.
Introducing Corcel
Corcel, formerly known as BitAPAI, was originally meant to provide API access to Bittensor sub-nets. However, it has gained popularity as a chat app that is just as good, if not better, than GPT for daily use. Graichen and his team are also working on incorporating vision into Corcel, allowing users to show it a photo and ask it to perform tasks such as cutting out objects or suggesting recipes.
The Bottom Line
When asked why someone should use Corcel over GPT, Graichen's simple answer is that it works better and it's free. While he acknowledges that GPT is the industry leader, he believes that Bittensor's open source models, which are often trained on GPT, are just as good. Corcel is currently free for consumers, and Graichen believes that revenue-generating apps will fund the network in the future.
Stay Ahead of the Curve with MarTech
The future of AI is constantly evolving, and Marcus Graichen's unique perspective sheds light on the potential for decentralized and open source AI tools. Keep up with the latest technological advancements and trends by subscribing to MarTech! Daily for free.
Originally reported by Martech: https://martech.org/marcus-graichen-looking-forward-with-ai/
This article was written automatically by artificial intelligence. Please make us aware if you have any concerns about this automatically generated content.
