This is a clickable demo. The video call is fabricated, and the call plays itself in about 45 seconds. Switch to After: new tab in the top bar at any time.
Ambient Answers is a prototype for "preemptive" search, a means to predict what a person is wondering about or does not know and provide answers before they explicitly ask a question. The prototype shows one capability in two moments. During a short fabricated video call about a zoning proposal, notes appear under the captions: terms explained, context added, and a correction when the other person gets a date wrong. Afterward, a new tab tells the user what they were wondering about, one question at a time, with the answer and sources. The call plays itself in about 45 seconds.
As answers become effortless, not knowing, an ordinary part of life, may come to be treated as a problem for technology to eliminate.
Anyone who uses AI systems in everyday life has this problem, as assistants, search and note-takers sit closer to every conversation. The time between meeting a question and getting an answer is approaching zero, and ambient systems could supply explanations before a person asks. These systems can be accurate, useful and deeply personalized.
It matters because the experience of not knowing, which prompts people to ask, wonder and look things up for themselves, could be designed away without anyone deciding that it should be.
Ambient Answers shows how AI systems that listen and participate could preempt search. They would consult a person's context, memories and cues, and read visual and linguistic signs of a lack of comprehension, to deliver answers before any question is asked.
The prototype shows this in two moments: in real time, annotating a call to reduce ambiguity, and afterward, as a brief of what the person was wondering about. Under that brief sits a quiet list of questions already answered for them. It is built around a fabricated video call, so the idea can be seen without any real person's data.
If a tool answers what you are probably wondering before you ask, during a conversation and again afterward, the gap between having an answer and getting to one becomes the thing to look at.
If that holds, the question of what the web is for shifts. It stops being where people go to find answers, and the thing to design for becomes what people still want to work out, wonder about or leave unknown.