Crypto Encyclopedia IQ.Wiki To Develop Own AI As It Upgrades To OpenAI GPT-4 Model

IQ.Wiki, a blockchain and crypto encyclopedia has upgraded to GPT-4 which is the latest model of OpenAI’s ChatGPT, IQ.Wiki’s community-led governing organisation, BrainDAO said on Thursday that it now uses the more advanced artificial intelligence technology available and will be focusing on developing its own AI with Wiki articles.

It was upgraded from the GTP-3 or the third-generation Generative Pre-trained Transformer which is a neural network machine learning model trained using internet data to generate any type of text.

IQ.Wiki used GPT-3 to summarise lengthy articles into 256-characters for search engines. The firm showed in a demonstration that GPT-4 produces a more precise and concise article than GPT-3.

Some of our pages like our wiki on the Bored Ape Yacht Club are over 5000 words. AI allows us to now summarise these wikis for search engines in seconds. This frees up time that editors can use to research, create, and edit, Andy Cho, director of Business Development at IQ.Wiki told Permaweb News.

Beyond shortening lengthy articles, Cho sees artificial intelligence as useful in making team members who do not speak the same language to collaborate easily. BrainDao said AI could help translate complex Wiki articles into dozens of languages, predict topics with potential to be popular and increase the utility of the IQ token.

The challenge with using traditional translation software like Google Translate to translate wiki articles is that words can have many meanings in different contexts, especially when dealing with specific industries or topics. The advantage of AI language models is that they can be trained for more specialised translation tasks and learn from feedback, Cho said.

However, despite artificial intelligence’s potentials and growing popularity, some have criticised it as posing a threat to jobs humans can do. But while Navin Vethanayagam, the chief brain of IQ.Wiki agrees that some jobs done by humans could be eliminated by AI, he believes training AI will create lots of opportunities.

Vethanayagam sees an opportunity for a collaboration between people and AI – where editors train AI models by rating and giving feedback to the language models that contribute content.

One of the challenges in building better artificial intelligence models is getting honest high quality feedback on AI responses, Vethanayagam told Permaweb News. “Specialised online encyclopedias like IQ.Wiki which have quality content that’s been reviewed by editors are a much safer place for AI models to get honest and useful feedback. As AI models improve based on editor feedback they’ll quickly be able to handle repetitive tasks like summarisation and spelling, grammar, and formatting edits”.