Say you once saw a youtube where a woman wet a paper towel and somehow put it on the tip of the dustpan so that when she swept the dirt into it, everything was neatly and cleanly gathered. Or you were on a website where you were able to download the driver and instructions for your BOQ19c which is a someThingorother that you don't even remember what it does, but do recall taking it out of the box.

B4 c.it you could ask google or GPT. Google would point you to quora, where somebody already asked this question but didn't get an answer, and to a big list of websites that seem to be what you are asking for, but after 3 hours you finally conclude that they aren't.

GPT or even the new Bing and Bard would answer confidently and without any doubt that you are talking about a movie from 1975 starring Peter Sellers. When you would contest that "knowledge" (presented to you in about 30 pages of info) and say that it was a recent youtube and not a movie it would apologize for the misunderstanding and point you to a children's book by Enid Blyton. Then apologize again and explain that it was not intended for this kind of task.

So how does c.it do.it better?

1. Homing in: It is a chat, asking you questions and listening to your answers leading you to the correct result, with much less wasted time. (So no 30 pages of information, and no 20-page lists of links, only short replies).

2. Immediate Knowledge Gathering: Rather than having any information in advance it searches the web in ever-changing circles as the discussion proceeds.

3. Analyzing the Results: And it READS THE RESULTS. Constantly checking if what it found is what you are looking for and if so, presenting it to you as a link with a short summary. (Up to 3 suggestions, and a headline or two for a different direction)

4. Follow up: So for example it looks up answers on Quora. Some of the answers include a link or two. Our c.it then goes to each link and (1) checks if it still works and (2) that it actually is what you want, or could help us lead you to what you want. It also checks what the responses were to this Quora answer and can tell you if it was considered by anyone to be a good answer, or perhaps didn't work for most of them.

5. Visual Aural and Verbal (Textual) interactivity: It actually looks up what it is recommending, including watching the videos or looking at the pictures and it then summarizes some of the main scenes if a summary is not already available for that media.

That way it could also visually discuss things with you, and play sounds that enhance the communication experience while having the verbal discussion with you in text.

6. Request for feedback. Hi, my friends! Was I clear this time? (biting my fingernails and waiting for the A1 flames)
-- pashute, Jun 23 2023

Total browsing archive Reminded me of this... [pocmloc, Jun 23 2023]

Text search in searched text ... which was inspired by this idea from [pashute] two-and-a-half years ago. [pocmloc, Jun 23 2023]

ok that was a bad example.

Googled "damp dustpan for sweeping" and got a short text explaining what to do, and a list of videos.

But the woman had the whole towel wet and I'm looking for one where only the tip of the towel was wet... She achieved that in a smart way. What was that?...

3 hours later... ok that WAS a good example.
-- pashute, Jun 23 2023


Yes.

1. Reviews AND SUMMARIZES its own results.

2. and asks for your feedback about its summary for each link for RELEVANCE and CORRECTNESS.

3. Short message conversational mode, with further options held back to be presented later in the discussion.
-- pashute, Jun 25 2023


hmm! thanks [pocmloc]
-- pashute, Jun 25 2023


//Visual Aural and Verbal //

Magic for now: requires effective multi-modal AI which does not yet exist
-- Voice, Jun 25 2023


which stage 1 to 5 is magical?
-- pashute, Jun 26 2023


//which stage 1 to 5 is magical?//

As I said, stage 5 is, because it requires effective multi-modal AI which does not yet exist. Hey, stop staring out the window. Recess is in just an hour. Actually if it just listens to the videos and runs them through text to speech it's possible, but loses a lot of context. But "visual" is right out.
-- Voice, Jun 26 2023



random, halfbakery