h a l f b a k e r yI like this idea, only I think it should be run by the government.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
It also stops when interrupted, and can correct itself. Please wait before you tell me how GPT-like AI won't work for this. I'm not talking about GPT-like AI. I'm seriously talking about my new Overstand technology which should be ready within the coming year...
Alarm 8:30!
- Set alarm today 8:30
pm?
No. 8:30 in the morning.
- Set alarm tomorrow 8:30 am? Or every morning?
Every morning except Shabbat.
- What is Shabbaz?
Its Shabbat. In Hebrew.
- Oh (writes in Hebrew) Shabbat! - Saturday? That makes sense. "Set alarm 8:30 am every morning except Saturdays." Correct?
Do it.
- I'm doing it. What is the alarm for, and which ringtone should I use? the last one was pulse.
No time now.
- OK. Just a sec. (I'll ask you about it again later)
- Hey Google! ...Set alarm 830 am every morning except Saturday.
- Your dictatorship is ready
Dictating: The Hamtu and Maru verbs in Sumerian are...
- just a sec. This is about the Sumerian language. Right?
Yes.
- ok continue. Meanwhile, I'm loading the Sumerian linguistics context.
OK, so the Hamtu and Maru verbs...
- Do you want me to repeat that out loud or just continue typing?
No just continue typing. I see what you're writing.
...(later, after 5 paragraphs) OK, important! Please Call Ron?
- Calling Ron. (if you're talking to me say Yup, if not say Nope) Meanwhile, the text and conversation are saved.
OK, now quiet till the lecture is over.
- (Whispering. Click sound. Shows thumbs up). Half hour reminder?
- (writing only) Oh I see on the calendar it's till 2:30. No problem.
Anybody here wanna join? This is an open invitation for my fellow halfbakers to contact me. pashute is my gmail.(it means simple in Hebrew, and is the name of my company). My real name is Moshe Flam but I got used to the halfbakery's [pashute] and even enjoy it!
Thanks [pocmloc]
https://www.youtube...watch?v=BETE4kVYmaU
[pashute, May 31 2023]
[link]
|
| |
Does it still work if you're not standing on a podium in a smart uniform in front of a bank of microphones? |
|
| |
This reminds me of P.G. Wodehouse's account of how he tried dictating into a recording machine instead of to a live secretary. |
|
| |
A useful (genuinely intelligent) AI personal assistant would be lucrative. There's a few companies developing them at the moment. If yours works and has a snappy name, who knows... |
|
| |
A Dictaphone that knows the difference between data and meta-data ... I think. |
|
| |
Join you in what, developing this? It should be possible in principle with many API calls to existing LLM's and very good speech to text. But I don't think LLMs are fast or smart enough, even if you have one iteration tracking the primary conversation and an iteration for each of the user's statements. I know you said your idea doesn't involve LLMs, but nothing else nonuhuman is smart enough to perform the rest of the duties. |
|
| |
Can I borrow your dictaphone?
No, just use your finger etc etc. |
|
| |
I sent an email the other day, but I haven't seen a response yet. Did you get my email, [pashute]? |
|
| |
yes. sorry, didn't look at emails in the past few days. |
|
| |
Join me in developing (in plan, theory and code) what I call Overstand technology. Instead of "getting it right", this technology knows its limits and uses the discussion with you to advance in steps toward the answer. |
|
| |
It can be interrupted in mid-thought, and in fact, does that all the time to itself when a better answer comes up in one of the competing threads. It's just that, unlike us, the interruptions don't distract it from what it was set out to achieve. |
|
| |
I creates a "lexicon and grammar" weighted network for each field of discussion, it has a summary and a mood and mode of conversation, even claiming to have feelings. It can be scientific or empathic, or funny. But this time it "gets" the joke before it tells one, and learns to categorize the humor and "see where it's going". |
|
| |
I am using the following types of information which all tie in to create artificial comprehension: |
|
| |
From linguistics: Phonetics, phonemics, accents and sounds. (Analyzing the sounds and finding out the Vowels and consonants, rhymes, imitations of nature etc). |
|
| |
Also lexical, morphological and syntactical analysis (which means finding the words in the dictionary, checking out the grammar that is used for those words, and analyzing the sentences and paragraphs that are built from it), |
|
| |
Semantics, pragmatics, context (getting the literal meaning along with several levels of possible extra or deeper meaning). Analyzing the style and goal of the topic discussed, and creating a "lexicon and grammar" for each kind of conversation and field of discussion. |
|
| |
We'll start like children with simple language, fulfilling the Overstands needs (which will perhaps be an artificial equivalent of some of the basic human needs). |
|
| |
It doesn't have to be fast (at least during the first stages). The three main things that make it different from the large models are: |
|
| |
1. that it listens to its own competing versions of thought before emitting the result, |
|
| |
2. That it remembers its conversations and can criticize itself on the go, before, after, and while emitting the answer. Reshaping its own thoughts. |
|
| |
3. And that the knowledge core itself is built comprehensively and not statistically so that it can tell you what it understood, and how it got to that understanding. It can correct its responses on the go and can show us where there are gaps and possible errors in its replies and thoughts. |
|
| |
Most of what you just said is just a description of how speech to text is done. Easiest to just send the text to Google or another service which already has all of it down pat. |
|
| |
No it doesn't. It does not learn the context that I'm speaking in, and if it doesn't know a single word and that word they way it thinks it got it doesn't make sense, it doesn't know how to ask me the correct questions so that it gets back on track. |
|
| |
If we discuss a topic which is not mainstream, or a foreign word, in the cases that it did not detect it correctly, it has no idea how to correct itself. |
|
| |
And the AI chats are even worse in this respect. They will go on and on in the wrong direction even when you try to point them in the right one. You either have to give up, or start afresh from a new direction. |
|
| |