How do startups prevent creating false messages the dead never said?

Question

Accepted Answer

Startups Employ Safeguards to Prevent AI from Fabricating Messages from the Deceased

Companies developing AI-powered digital replicas of deceased individuals are implementing strategies to ensure the generated content remains true to the individual's life and validated communications. This involves a focus on using existing recordings and data, rather than allowing the AI to create entirely new, unvalidated statements. The aim is to capture the essence of a person's interactions and memories without venturing into fabricating potentially distressing or inaccurate messages.
Prioritising Existing Validated Content

Joseph Murphy, head of development at DeepBrain AI, has explicitly stated that their "Rememory" programme is designed to avoid generating new content that the deceased would not have uttered or validated during their lifetime. This principle guides the development of their digital replicas, ensuring that the AI operates within the bounds of the individual's recorded or written communications.
Capturing an Individual's Persona

StoryFile, another company in this space, focuses on capturing the unique qualities and experiences of an individual. Their approach involves gathering extensive material that represents the person, which is then enhanced and utilised with AI tools. This method aims to preserve the authenticity of the individual's presence and communication style.
User Trust and Control as Key Concerns

While the technology allows for the creation of sophisticated avatars, companies acknowledge the sensitive nature of this work. There is an understanding that users, particularly those experiencing grief, need to trust the AI's output. Concerns about AI potentially saying something unacceptable or the user feeling a lack of control over the avatar's actions are significant factors that influence the design and implementation of these digital replicas.

Reading history

Hide →

This history is cleared when the page is closed.

How do startups prevent creating false messages the dead never said?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends