Step-by-Step Guide to Using Personal AI
Whether it's future dinner plans with friends or sharing your expertise with your co-workers, everything is indexed automatically and is entirely discoverable simply by chatting.
Personal AI isn’t pre-trained with an algorithm for a given dataset. In fact, it does not even import its own data into your AI. This is to keep your personal AI neat and pristine with your purpose, with no other organizational agendas. The goal is to increase your personal scores over time - to evolve your AI so it becomes an extension of you. Messaging your AI like a one-time chatbot will leave you disappointed. It's an investment.
When you ask questions your AI has no knowledge on, you will receive a concise general response from GPT-3 (Open AI) with a very low personal score. Our intention is that we want to provide your AI with a foundation of simple facts. Personally, I have no opinion on Elvis, and I would most likely say something general like the response below. Focus on getting data in and messaging your AI about the imported data, and training your AI about what matters to you.
General AI response with low personal score
Tip 📚: To understand the model and how quickly it train’s, send a basic message or bio about yourself, stack it, and ask about it (ex: my name is….I am from…my job is….).
Then ask, what is my name? This example may be vanilla but it is a representation of the bare bones of the model. It teaches us how quick the model learns. It teaches us how to structure data given how we or others will ask questions (ex: if you send an autobiography written by someone else in first-person, the AI will not be able to differentiate you vs. the author).
Think through your use case and about the data you have. What viable and personal text, voice, or video content to you have? If you were teaching someone about yourself what would you tell them? What do you want your AI to be able to answer? Who do you want to share your AI with?
The use cases are endless, ranging from developing a personal knowledge management system to setting up an AI for a wedding, which can handle all the logistical details and answer guests' queries. People often ask how do I teach by AI about X or Y. You control the data and model. It will act on the data you input. Do not hesitate to share and receive advice on your use case in community.personal.ai.
Tip 📚: We suggest everyone to start to message contacts as soon as possible in Personal AI because it allows your AI to learn about you daily without heavy lifting. Your AI learns from your messages, and if you use personal AI as a communication medium, your AI will receive more data on you. If your AI is for clients, messaging clients will teach your AI more about your company and how you speak (ex: formal). If it is for your professional life, your AI will be up to date on your projects because conversations with your team will be continuously stacked. If it is for your personal life, talking to your family and friends will allow your AI to be up to date on events, memories, and how you communicate in your personal life.
What are you hoping to add to your personal knowledge? What do you want to learn? What are your own goals or KPIs? For example, I work in consumer AI, but I have no personal content, blogs or articles on consumer AI. My goal would be to consume several articles, blogs, and videos on the topic and add it to my knowledge.
Tip 📚 : If you want to keep external data separate from your own content, data, and memories, upload external data in third-person. Therefore if I was uploading a section of Gandi’s autobiography, I would change the letter ‘I” to “Gandi” by pressing Command F in Google Drive OR by asking ChatGPT to rephrase it to third-person.
(Important note: All profiles get merged into the main AI. 🥣)
Think through the various parts of your life, and with AI profiles, you can separate different parts of your life. For example, I do not want to share my personal memories about my friendships with my professional AI. We all want to share our AI with our contacts confidently and assess what we want to share. Therefore optimizing your AI’s early on is effective. Furthermore, when messaging your own AI it is helpful to control the query of data being used. For example, I once uploaded Star Wars movies to create a Star Wars AI for a lounge with Star Wars enthusiasts. I quickly realized the importance of separating data queries. Whenever I messaged my AI, it started responding like Luke Skywalker - "The force is strong with this one" or "I have a bad feeling about this". Now I separate external and internal knowledge.
We all have our own value, priorities and goals. Sometimes they may be held in our minds, or a notebook deep in our drawer. Train your AI on what makes you tick to ensure that it is a true reflection of yourself.
a. Your preferred content, goals, memories, values, and beliefs. 🥇
b. Content you enjoy and find valuable. 👨👩👧👦
c. Your communication style and preferences. 📬
d. In-depth knowledge and resources related to your areas of interest or expertise. 🎓
e. Know the data in your memory stack so you know what to ask. 📝
a. Compile relevant information from your memory stacks. 📚
b. Help with answering questions and generating content using built-in GPT. 🤖
Yes, your AI can answer questions similar to ChatGPT prompts
c. Analyze patterns in your content and create new ideas with your content. 🎨
d. Continually update and expand your knowledge vault. 🔒
e. Draft messages, emails, and other forms of communication. ✉️