re: The OpenAI Chaos this weekend may have been about an AI breakthrough afterall

quote:

---

Though only performing math on the level of grade-school students, acing such tests made researchers very optimistic about Q*’s future success, the source said.

---

What kind of computers are not capable or performing simple arithmetic? This sounds like nothing to me

Man will be controlled by AI sexbots... so not much different than now.

AI can be the God we all wanted.

We always wanted someone to pay an immediate price when they broke the law, AI can easily be that “lightning bolt” from the sky that strikes down sinners where they stand.

Drones with tasers, facial recognition, AI control.

You just jaywalked? ZAP

You just littered? ZAP

It’s your second time today? ZAAAAAAAAAP

There's lots of speculation about what this breakthrough could possibly be, but based on the name and the rumored achievement, most believe that they've built an AI that is much better at reasoning its way through problems. The current LLMs rely very heavily on their training data and often fail when asked to step outside of that data. That's why they are so bad at math for instance, and examples of this one being good at math suggests a leap in ability.

Here's one of the better explanations of what this might be that I've seen so far:

quote:

---

Your First AI @YourFirstAI

If Q* is what I think it is… this is absolutely BONKERS

Q-learning has existed for decades already. It’s just a basic reinforcement learning algorithm. A* is also fairly old- it’s a heuristic-based path finding algorithm.

In typical engineering fashion, they may have found an intersection of the 2 and named it Q*. This is total speculation, but if this is a “breakthrough” that means OAI built an algorithm that can feed a highly efficient heuristic into Q-learning. That is MASSIVE.

I’ll save the boring details- what does this mean in real terms?

Learning is a long path: a machine must accomplish many small steps to achieve a larger task, and if those steps are not pre-determined, the machine will try many combinations of steps to achieve a goal. Reinforcement learning will “reinforce” the optimal steps to bring a machine closer to its goal. Think of a child trying to walk- he may fall over many times as he tries to find his balance.

A heuristic is a measurement that a machine uses to assess success. As you change and improve a heuristic, a machine will have a better assessment of success.

What Q* might have done is bridged a big gap between Q-learning and pre-determined heuristics. This could be revolutionary, as it could give a machine “future sight” into the optimal next step, saving it a lot of effort. This means that machines can stop pursuing suboptimal solutions, and only optimal ones. All the “failure” trials that machines used to have (eg. trying to walk but falling) will just be put into the effort of “success” trials. OpenAI might have found a way to navigate complex problems without running into typical obstacles.

There is so much good research out there. If you’re interested in learning this, Q-learning and A* are both highly documented and well-researched, and a component of most university-level CS curriculums.

In the last few years, research teams have been trying to bridge the two using hyper-heuristics, you can Google those as well. Good luck as this is a deep rabbit hole.

---

AGI won’t stop or prevent the 100% off sales going on in California retail

Can you link that post?

quote:

---

Can you link that post?

---

LINK

quote:

---

I’m teaming up with AI if they ask. Humanity had its shot

---

Logical AI will be based af. No DEI bullshite. I’m team AI.

quote:

---

What kind of computers are not capable or performing simple arithmetic? This sounds like nothing to me

---

This isn’t a “computer”. This is a program, only unlike most other programs where you build the logic of things in like “If This Then That”, these AI programs are designed to learn like humans. The consume large amounts of data and store patterns so it is very good at predicting things like “if someone asks A, the answer that they should get is B”. That’s how the visual models work too where you tell it to draw something and it does.

Back to the math. These models are built on pattern recognition so if something isn’t in its data set it becomes a lot worse at predicting the answer. This alleged break through is believed to help the program reason its way to an answer rather than relying so heavily on pattern recognition.