For people willing to use AI after losing someone close, the appeal often centers on unfinished communication. Their responses also show clear limits around how much an AI system should be allowed to invent on behalf of someone who has died.
The idea of recreating a deceased person with AI is often framed around simulation: whether software can reproduce someone’s voice, manner or likely responses closely enough to make an interaction feel familiar. The more immediate question is what grieving people would actually want to do with that technology if it were available to them, and how much authority they would give it to speak for someone who is no longer alive.
To examine those preferences, Use.AI surveyed 10,800 adults who had experienced the death of someone close across the United States, the European Union, the United Kingdom and Latin America. 54% said they would consider having a single AI-generated conversation with that person. The remaining findings refer to respondents who said they would consider having such a conversation.
Among those respondents, 46% said there was something specific they still wished they had told the person they lost. 42% said they would use the conversation to tell that person what had happened since their death.
Asked to choose the single main purpose of the encounter, 29% said they would most want to say something they never got to say, the largest group. 21% would mainly want to share what had happened since the person died, while 17% would most want to hear something they had wished to hear while the person was alive.
The pattern suggests that much of the interest is connected to communication left incomplete by death. Respondents were often focused on having somewhere to direct their own words, rather than simply receiving new material generated in the deceased person’s voice.
That preference also appeared in the format people chose. 44% preferred a one-way experience in which they could speak to the person they lost without the AI generating new words on that person’s behalf. 38% preferred a two-way AI conversation.
Responses became more cautious as the system was given greater freedom to generate new material. 24% would limit the AI to words the person had actually said or written. 26% would allow it to rephrase views the person had genuinely expressed, and 25% would allow it to infer what the person would probably say. Only 10% would allow entirely new responses designed to fit the person’s personality. 15% said AI should not speak on behalf of the dead at all.
A similar limit appeared when respondents were asked about emotionally comforting responses with no evidence that the deceased had believed or intended them. 41% said they would not want the AI to generate such words. 29% would accept them only if they were clearly identified as AI-generated, while 18% would want the comforting response regardless.
Existing records of the person were also valued more highly than newly generated conversation. 45% said they would rather have access to real messages, recordings, photos or videos than a new AI-generated exchange. 27% said both would be equally meaningful. 18% preferred the AI conversation.
Interest in a conversation also did not automatically extend to permanent access. 35% said they would prefer one limited AI conversation rather than an ongoing simulation. Asked what should happen after a one-time encounter, 34% said they would want the simulation permanently deleted. 37% would keep it available only for deliberate reactivation, while 18% wanted permanent access.
The survey points to a relatively constrained use of this technology among people who are open to it. Many respondents were interested in completing communication that death had interrupted, while support fell as AI was given more freedom to generate new speech or maintain an ongoing version of the deceased.
AI cannot establish what someone who has died would have chosen to say. For many respondents, the more acceptable role was narrower: providing a setting for unfinished communication while preserving clear boundaries around what the system could add on its own.
About Use.AI
Use.AI is a universal AI assistant that aggregates the world’s leading large language models into one unified and seamless experience. It provides users with a single point of access to the most advanced AI capabilities available today, from complex problem-solving to creative content generation. By bridging the gap between multiple AI technologies, Use.AI empowers users to enhance their productivity and leverage cutting-edge intelligence in their daily workflows.
Media Contact
Alex Samuels
PR Manager
Use.AI
pr@use.ai
To examine those preferences, Use.AI surveyed 10,800 adults who had experienced the death of someone close across the United States, the European Union, the United Kingdom and Latin America. 54% said they would consider having a single AI-generated conversation with that person. The remaining findings refer to respondents who said they would consider having such a conversation.
Among those respondents, 46% said there was something specific they still wished they had told the person they lost. 42% said they would use the conversation to tell that person what had happened since their death.
Asked to choose the single main purpose of the encounter, 29% said they would most want to say something they never got to say, the largest group. 21% would mainly want to share what had happened since the person died, while 17% would most want to hear something they had wished to hear while the person was alive.
The pattern suggests that much of the interest is connected to communication left incomplete by death. Respondents were often focused on having somewhere to direct their own words, rather than simply receiving new material generated in the deceased person’s voice.
That preference also appeared in the format people chose. 44% preferred a one-way experience in which they could speak to the person they lost without the AI generating new words on that person’s behalf. 38% preferred a two-way AI conversation.
Responses became more cautious as the system was given greater freedom to generate new material. 24% would limit the AI to words the person had actually said or written. 26% would allow it to rephrase views the person had genuinely expressed, and 25% would allow it to infer what the person would probably say. Only 10% would allow entirely new responses designed to fit the person’s personality. 15% said AI should not speak on behalf of the dead at all.
A similar limit appeared when respondents were asked about emotionally comforting responses with no evidence that the deceased had believed or intended them. 41% said they would not want the AI to generate such words. 29% would accept them only if they were clearly identified as AI-generated, while 18% would want the comforting response regardless.
Existing records of the person were also valued more highly than newly generated conversation. 45% said they would rather have access to real messages, recordings, photos or videos than a new AI-generated exchange. 27% said both would be equally meaningful. 18% preferred the AI conversation.
Interest in a conversation also did not automatically extend to permanent access. 35% said they would prefer one limited AI conversation rather than an ongoing simulation. Asked what should happen after a one-time encounter, 34% said they would want the simulation permanently deleted. 37% would keep it available only for deliberate reactivation, while 18% wanted permanent access.
The survey points to a relatively constrained use of this technology among people who are open to it. Many respondents were interested in completing communication that death had interrupted, while support fell as AI was given more freedom to generate new speech or maintain an ongoing version of the deceased.
AI cannot establish what someone who has died would have chosen to say. For many respondents, the more acceptable role was narrower: providing a setting for unfinished communication while preserving clear boundaries around what the system could add on its own.
About Use.AI
Use.AI is a universal AI assistant that aggregates the world’s leading large language models into one unified and seamless experience. It provides users with a single point of access to the most advanced AI capabilities available today, from complex problem-solving to creative content generation. By bridging the gap between multiple AI technologies, Use.AI empowers users to enhance their productivity and leverage cutting-edge intelligence in their daily workflows.
Media Contact
Alex Samuels
PR Manager
Use.AI
pr@use.ai