A Brazilian technology promises to create a digital version of individuals with a high degree of personalization. Deep Reflection is based on the connection between thought patterns, opinions and communication style to contribute to the creation of content and the work of digital influencers.
The big difference in relation to other existing AI solutions is the broad domain of a specific area of knowledge. The solution deals with extremely specific content and can mimic aspects of individual consciousness, reinforcing your logic, worldview and the way of interpreting information.
Artificial intelligence learns through different text models, videos and images and, in this case, it even replicates the rhythm and way of speaking. According to the co-founder and researcher at Vortice.ai, Giovanni Laporta, technology creates “digital reflections”: AIs trained based on a single person’s content, as content creators, teachers or artists.
Minor models, bigger focus
This approach differs from the path followed by big techs, who invest massively in the production of LLMs (Large Language Models), like ChatGPT or Gemini. In the case of Deep Reflection, the bet is on SSMLs (Small Language Models), small models trained with specific knowledge for a particular person or situation.
This training process is considered faster, as Deep Reflection's audience involves a reduced mass of data when compared to the volumes used in LLMs. “Instead of me training a great model who knows everything in the world, I will train small models, and each small model has specific knowledge”, explains Laporta.
At the moment, Deep Reflection is mainly used by content creators, allowing the public to interact with this digital reflection through the same communication model, including words, original expressions and jargon. However, technology can be applied to different sectors, while raising discussions about the risks of its misuse, such as deepfakes and the loss of copyright traceability.
