Living with My AI Twin (I): The Meaning of Identity and Education in the Age of Personal AIs

On March 15, 2024, I introduced the world to my AI Twin — “Paul AI”. What exactly is the purpose of “Paul AI”? How do I interact with my digital counterpart, and how do others engage with it? Check out this post to discover more and learn how my life, the life of real Paul, has transformed since the birth of Paul AI.

Hello, “Paul AI”

Have you ever tried to imagine a world where you could have your own digital twin — your own digital counterpart that could interact, learn, and even teach on your behalf? I knew that world wasn’t far off, but I have never thought it could become possible so quickly.

March 15, 2024 marks the day of “Paul AI” — it is my own digital twin that serves as a gateway for my friends, social media followers, my business partners and students to interact with my areas of expertise in a more discreet manner.

How does “Paul AI” work?

“Paul AI” is built on top my publications — dozens of law review articles, blog posts, and several books as well as my conference talks and webinars.

This knowledge base resides in my personal data cloud and is knowledgeable about topics I have been exploring in nearly 15 years of my professional career such as copyright, intellectual property and international law, and more recently, data privacy and generative AI.
​
What “Paul AI” actually does, it provides a way for anyone a way to interact with my specific areas of expertise — and search my knowledge based in a new way.

Thanks to the advancement of AI and large language models, search has become a commodity. By having most of my publications and conference talks in one place, I can interact with such knowledge base by typing in any question, and get quite a solid response.

You can try interacting with my digital AI twin — “Paul AI” — here and experience it yourself.

What about security of data?

Data security is one of the most paramount issues. In the case of “Paul AI,” we utilized publicly available information — thousands of pages of published law review articles, recorded conference talks, podcast interviews, and blog posts.

Previously, one would have had to search online and read through this content to find an answer based on my area of expertise. However, in the age of AI, thanks to new data search capabilities, “Paul AI” offers the speed and previously unimaginable opportunities to interact with data in a new and dynamic way.

Since the components of “Paul AI’s” knowledge base come from public sources, there is very little concern about data security: this information is already available on various platforms and databases.

Moreover, as the creator of “Paul AI,” I have full control over what data sources constitute the knowledge base. I can add more sources if I find that necessary or remove some that are currently included.

In addition to having full control over the knowledge base, I can also benefit from the dashboard to see what conversations are taking place with my AI Twin: what questions are people asking? How is Paul AI responding? However, I cannot see the identities of the people interacting with Paul AI.

Paul AI: Today and Tomorrow

​As of today, the primary function of “Paul AI” is to provide a new way to interact with my specific knowledge base. This means you can search my knowledge base by typing in text and interacting with “Paul AI” via text.

But what about using your voice? Can “Paul AI” respond in the real Paul’s voice? You might wonder.

Indeed, numerous companies already offer solutions to clone your voice and likeness in a matter of seconds.
​
At Prifina, we are tackling the more challenging task of allowing users to build a knowledge base from various data sources. These sources can include not only text, as in the current version of Paul AI, but also other types of data. Imagine incorporating heart rate data from your fitness tracker, or your medical records.
​
Future versions of Paul AI will not only speak in my real voice but will also feature a virtual body avatar that reacts correspondingly.

Furthermore, I could create additional twins — like “Paul Health AI,” which would hold not just my medical history but also recent biometric data, such as information about my activity and sleep patterns. Who might be interested in interacting with “Paul Health AI”? Potentially my family doctor, my basketball coach, my nurse, or even my life insurance company representative.

The possibilities and opportunities for Personal AIs are immense:

First Use Case: Teaching at Vilnius University Law Faculty with “Paul AI”

I am particularly excited to deploy “Paul AI” as my AI assistant for the Data Privacy Law class I began teaching last week at Vilnius University Law Faculty.

This year, I’ve decided to go “all in,” requiring students to utilize all available AI tools for class preparation and assignment completion. This includes #ChatGPT, #Claude, Meta’s AI, Inflection AI’s #Pi, Perplexity , and, of course, Paul AI. Each of these tools has its own pre-programmed limitations, and I aim for students to learn how to effectively collaborate with them.

In my course, titled “Data Privacy Law: Human-Centric Approach,” “Paul AI” offers specialized knowledge on the subject. Over the past five years, I have extensively explored the human-centric approach to personal data — a topic that other popular AI assistants do not seem to cover as comprehensively.

​It has become clear, that “Paul AI” offers me an opportunity to curtail the duration of office hours, and reimagine the way how I organize the class, interact with the content and teach law-related subjects in such a rapidly evolving environment.

“AI is a gift to humanity that has been largely misunderstood.” — Prof. Mark Fenwick

There are many ways AI can be applied in the classroom beyond just having students write essays using AI chatbots. While the internet has already broadened access to education, some may be wary of introducing AI into the classroom. However, it’s undeniable that artificial intelligence is set to transform education.

The potential to further incorporate AI is substantial. It’s estimated that teachers spend 40% of their time on tasks such as grading and course planning. AI can assist with these essential but repetitive tasks, freeing up more time for professors to develop relationships with their students.

AI can also enable students to learn faster and access a higher quality of education. While many consider this the future of education, it’s actually already happening right now!

What do you think?  We’d love to hear your views. In the future, we’ll continue sharing my experiences and exciting opportunities with personal AIs — stay tuned!

Want to create your own AI twin?
Sign up here — powered by Prifina — Liberty. Equality. Data.