Posted on 07/15/2024 1:15:48 AM PDT by Jonty30
OpenAI’s rumored project Strawberry is focused on enhancing the reasoning capabilities of its AI models. Strawberry is rumored to be based upon the QStar AI advance. QStar was a reasoning advance that supposedly triggered the power struggle inside OpenAI and temporary firing of CEO Sam Altman.
Strawberry supposedly enables AI to perform long-horizon tasks (LHT), which require a model to plan ahead and execute a series of actions over an extended period of time. This technology aims to improve AI’s ability to understand and interact with the world, allowing it to autonomously search the internet and conduct deep research. By improving how AI models think and plan, OpenAI hopes to unlock new levels of intelligence and bring AI closer to human-level intelligence or beyond.
ChatGPT maker OpenAI is working on a novel approach to its artificial intelligence models in a project code-named “Strawberry,” according to a person familiar with the matter and internal documentation reviewed by Reuters. The project, details of which have not been previously reported, comes as the Microsoft-backed startup races to show that the types of models it offers are capable of delivering advanced reasoning capabilities.
How Strawberry works is a tightly kept secret even within OpenAI, the person said. The document describes a project that uses Strawberry models with the aim of enabling the company’s AI to not just generate answers to queries but to plan ahead enough to navigate the internet autonomously and reliably to perform what OpenAI terms “deep research,” according to the source.
I think this technology potentially a boon if it were used as an educational aid to teach people, like a rudimentary holodeck. I think kids could advance their education at least twice as fast as they currently can under an educational aid like AI.
The problem is that many people don't want to learn.
In fact, we already have plenty of resources for people who really do want to learn (including AI at its current state and resources like Khan Academy) but some people are lazy and backward-looking.
Even in its current state, AI is going to divide humanity into 2 groups (at a minimum): First, those people who want to learn and get smarter using AI as a tool, and second, those people who hate exercising their brains and are quite happy to remain stupid.
Those in the second group (many of whom are currently in college🤣) are about to become useless eaters and in doing so will imperil themselves.
I think people do want to learn, but our education system makes learning so little fun that people choose to be dullards instead.
However, I think that if people could see fruits from their learning much sooner, they’d be more motivated.
Those who control AI will continue to push the boundaries of knowledge with the expressed purpose of putting every one of us into the useless eater camp. Even if you are in the top 1%, eventually you will be there as well. :)
As they are currently structured, the AI platforms I use do not learn from their interaction with me. Their learning is fixed and intentionally anchored in their "training" data sets.
Soon (I hope), you and I will be able to subscribed to an AI platform that can learn along with us and adapt to our own personal ways of thinking and learning.
I think personalized RAG (Retrieval Augmentation Generation architecture) is a step in that direction, but I haven't built a RAG for myself...yet.
Amen!!!
The first victim of the school system is imagination. Children are forced to view learning as an unpleasant but necessary task.
Think how exciting school would be if children were introduced to the exhilarating life-long process of exploration of the unknown (which is what learning really is). The adventure never ends.
This is true but you can AI knowledge with private data. So if you are able to capture the interactions in a meaningful but static way you may be able to avoid some of the issues you are having.
It does come at an increased cost depending on how much data you need to capture to make this work. You would also need an upscale account with your favorite AI vendor.
Medium has a bunch of articles on how to customize ChatGPT with private data.
The ChatGPT store has some private/new data models constructed by individuals that you can use. You can even create your own though I haven't delved into the details. The ones I have looked at in the ChatGPT store are essentially black boxes and I don't see any information as to how they were trained on specialized data.
I am a ChatGPT plus subscriber.
The problem I mention is not so much one of data as it is one of getting an AI platform to adapt to me (to recognize me and record what we have learned together as part of its training).
This is most notable when I have spent a great deal of time working with ChatGPT on a project. From one day to the next, it forgets what it might have learned from the prior day. It has a record of our prior discussions but what is learned (concepts and principles) from those discussions doesn't get included in its knowledge base.
I will check out the ChatGPT store.;-)
I call this the Age of Unreason. Never in human history have we had so much information available. Thanks to the left most of it’s wrong.
didn’t Ronald Reagan say something along the lines of democrats know so much that isn’t so?
I listen to You Tube videos often. I’m sure they use AI. One was about a ship and it referenced a ship’s bow often and pronounced it correct almost every time. But a couple of times it said bo instead of bow. Just one example but I still think that it’s not ready for prime time.
Kind of like in 2009 when TV switched from analog to digital TV. They weren’t ready but did it anyway. They eventually worked it out but it was still annoying. That’s where I believe we are now with AI.
It doesn’t have to do everything to be devastating. It only needs to do most of what we can do and most of us will be unemployable, because the work left over will not be enough to give everybody.
Here ya go - becoming proactive.
ChatGPT maker OpenAI is working on ... an artificial intelligence models in a project code-named “Strawberry,”
= = =
Watch out when it starts calling itself “Huckleberry”.
We are so screwed.
Slowly edging our way to SkyNet.
As long as it is limited to searching the internet it will be a dunce.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.