Integrating AI in Recreating Iconic Personalities

Deepfake Technology

OpenAIAPI

Elevenlabs

Project Overview

Artificial Intelligence is transforming how we understand and engage with the world, and history is no exception. This project explores the intersection of AI technologies—language modeling, voice synthesis, and animation—to create an interactive digital version of President John F. Kennedy.

Our team set out to prototype what it might look like to hold a conversation with a historic figure. We chose JFK due to the extensive archive of audio, video, and writings available, as well as his cultural familiarity. The goal was to create a believable and expressive simulation that evokes not just how JFK sounded, but how he thought and interacted.

👉 View our demo here
👉 View the original project pitch

Technical Stack & Implementation

OpenAI Integration for Personality Emulation

At the heart of the project was a custom OpenAI assistant. I used JFK’s public speeches, interviews, letters, and secondary sources (biographies, news coverage, etc.) to shape the assistant’s tone, diction, and ideological leanings. Responses were prompted to stay grounded in this curated corpus to reduce hallucination and better emulate historical plausibility.

Voice Synthesis with ElevenLabs

To add auditory realism, we integrated ElevenLabs’ voice synthesis platform. Using archival audio, we constructed a vocal model. This model could articulate any given text in JFK’s distinctive voice.

Digital Portrait & Visual Representation

In parallel with the voice synthesis, I created a digital portrait of JFK that could animate in sync with the AI’s responses. The goal was to provide not just a voice but also a face—an avatar capable of emoting, blinking, and displaying expressions. This combination of auditory and visual elements allowed for a more immersive and human-like interaction.

Toward Full Interactivity: Speech Recognition

While the demo relied on preloaded questions, a fully realized prototype would require real-time speech recognition. This would allow users to pose spontaneous questions and receive contextual responses. Integrating speech-to-text with the assistant and ElevenLabs voice synthesis remains the next logical step in development.

Challenges & Innovations

Two major challenges emerged:

• Historical Accuracy: AI models are prone to hallucination, which is especially problematic in projects claiming to represent real people. To mitigate this, I constrained outputs using a curated set of JFK’s speeches, writings, and reputable secondary sources. Even so, perfect fidelity is impossible—there are limits to what we know, and inevitably some interpretive liberties enter the simulation.

• Multimodal Integration: Getting different technologies—LLMs, voice synthesis, video animation, and (eventually) speech recognition—to work in unison proved difficult. This project was an early attempt, but a fully seamless system would require significant engineering effort.

Educational & Cultural Impact

This project points toward new ways of engaging with history. Instead of passively reading a textbook or watching a documentary, students could “converse” with historical figures, asking them questions and receiving responses modeled after their documented views and style. While such conversations will always involve a degree of artifice, the interactivity makes history feel more alive and accessible.

History is often presented in a cold, detached manner. By reintroducing voice, personality, and presence, AI opens a more experiential mode of learning. Imagine classrooms where students debate with simulated philosophers, activists, or world leaders. Even if imperfect, such tools can spark curiosity and deepen engagement.

Looking Ahead

This JFK prototype is just a beginning. Future work would involve:

• Real-time speech recognition for open-ended interaction.
• More advanced facial animation synced to emotion and tone.
• Broader historical libraries, allowing students to interact with multiple figures.
• Ethical guidelines for accuracy, representation, and misuse prevention.

As AI technologies advance, projects like this invite us to think critically about how we want to remember, represent, and interact with the past.