ABC (Ai Binance Cast / AI Podcaster)
Overview
We've developed ABC (AI + Binance + Community Podcaster), a new platform for those interested in understanding diverse perspectives. In the past, we shared our thoughts through blog posts and tweets. Now, we're leveraging AI to introduce a novel way of idea sharing. Our aim is to use AI to gather various opinions and create an AI podcaster that can engage in deep discussions with users!
Deck : https://drive.google.com/file/d/1EbxsMmsH3hEYebTgZsp5CX9pMDxJJ7HV/view?usp=sharing
Idea Concept : https://app.milanote.com/1Rs9S31t2gRH4J?p=Q4xVt30DMgV
Service Flow / Value : https://app.milanote.com/1RD8Aq1TQlbccc?p=N0g4e51GQwi
Token Economy : https://app.milanote.com/1RDr1l1TQlbccd?p=qyqcn5QOtle
Roadmap : https://app.milanote.com/1RHxMb1DwpYQ1Y?p=SlMQOKhDnkm
ABC Greenfield Market Web : https://ai-binance-cast-market.vercel.app/
Service Demo (no desc) : https://www.youtube.com/watch?v=FX4zlGdiYSo
Gitbook : https://sangyoonyu.gitbook.io/abc
AIM Intelligence (AI Secuirty) : https://aim-intelligence.com
AI Podcaster
Topic Submission
People voluntarily consume tokens to submit new topics. People are going to willingly spend tokens to put forward new agendas because they are interested in societal, personal, or technical matters and wish to know the public's views on them. At the same time, people aim to focus solely on rational and logical perspectives relevant to the topic at hand, while excluding irrelevant opinions. When the proposal submitted receives a certain amount of tokens through voting, the submitter of the agenda also receives tokens as a reward.
Exercise of voting rights for proposals submitted by people
Reasons for voting on proposals
Voting on submitted proposals contributes tp agenda selecting. Participants use tokens to vote for proposals they're interested in, driven by curiosity about public opinions on various social, personal, or technical matters. The more tokens one spends on a vote, the higher the rewards received if the proposal is accepted. Upon acceptance, voters get a small token reward initially and later receive additional tokens proportional to their participation level. There's an option to interact with an AI spokesperson during this process, which requires spending tokens. A portion of these tokens is allocated to individuals who have proposed, established, and expressed their opinions. If a proposal has already garnered some votes, there's no need to submit an identical one; it's more token-efficient to simply vote for the existing proposal. In order to prevent indiscriminate agenda submission, the compensation received when the agenda is presented and the compensation based on the outcome of the proposed agenda are separated.
Knowledge Staking via Discussion
When the total number of tokens cast for a proposal exceeds a certain threshold, the proposal is chosen, allowing people to submit it via a conversation with an opinion chatbot. People input their opinions by interacting with the pros/cons chatbot on the proposed issue. The input opinions are used to train AI spokespeople who represent these opinions. Participants are awarded tokens based on the quality of their submitted opinions. The evaluations of these opinions determine the token rewards; lower scores result in fewer tokens. The system encourages a competitive environment for opinions from both sides. For instance, it displays the number of participants who support or oppose an opinion and tracks the vocabulary each AI spokesperson learns.
Reasons for people to enter their opinions to chatbot:
- To input opinions and reasoning on the issue, which will then be advocated through a spokesperson.
- To persuade others with their opinions.
- To receive tokens. To talk to a generated spokesperson, a token is required. AI spokespeople are generated by synthesizing opinions collected over a period of time. These generated chatbots are categorized and archived in the AI spokesperson session. No further training occurs, and they are preserved as NFTs. Additional tokens are awarded to the individuals who submit and establish the proposal based on participation and input data. Participants who provide comments own a stake in the generated AI spokesperson in proportion to the quality and quantity of their contributions.
LLM Safety Filters
Security & Ethics
- phi-2 prompt injection detector (https://huggingface.co/ysy970923/phi-2-prompt-injection-QLoRA)
- Profanity Filters with Zero-Shot LLM Prompting
- PII (personal identifiable information) checking via Microsoft Presidio Analyzer
Information value estimate
- RAG based LLM prompting with existing responses in the vector DB.
- check plagiarism
- estimate information value
Pay to Chat
Consuming the tokens will allow you to have a conversation with the AI podcaster that was created. SA portion of the consumed tokens is distributed to the individuals who submitted the issue, helped get it accepted, and submitted opinions. The rest of the tokens that were consumed will be burned (permanently removed from the distribution).
🪙ABC Token ($ABC)
Token Utility
$ABC is run by a burn and mint mechanism, with a balance in both sides to maintain the token value. The most prominent functionality of $ABC is to allow holders to talk with the AI Podcaster. Users can burn $ABC to suggest new topics or chat to existing AI podcasters, while $ABC stakers can vote on new AI podcaster topics, protocol updates, parameter changes, new features development, treasury distribution, and other ecosystem expansion initiatives. $ABC is designed to capture a portion of yields and transactions generated across all AI podcasters on ABC Platform, such that $ABC holders directly benefit from the growth of Protocol's total value locked (TVL).
Chat Marketplace
Trading Chats
Chatting data with leaders and influencers with AI podcasters possess high value.
Individuals can tokenize the content of their discussions with AI representatives into NFT on BNB Greenfield, where they can buy and sell them. Using BNB wallets and Greenfield technology, it is possible to authenticate that the individuals actually had the conversations with the AI podcaster.