Lugh

joined 2 years ago
MODERATOR OF
 

Alibaba's Qwen team just released QwQ-32B-Preview, a powerful new open-source AI reasoning model that can reason step-by-step through challenging problems and directly competes with OpenAI's o1 series across benchmarks.

The details:

QwQ features a 32K context window, outperforming o1-mini and competing with o1-preview on key math and reasoning benchmarks.

The model was tested across several of the most challenging math and programming benchmarks, showing major advances in deep reasoning.

QwQ demonstrates ‘deep introspection,’ talking through problems step-by-step and questioning and examining its own answers to reason to a solution.

The Qwen team noted several issues in the Preview model, including getting stuck in reasoning loops, struggling with common sense, and language mixing.

Why it matters: Between QwQ and DeepSeek, open-source reasoning models are here — and Chinese firms are absolutely cooking with new models that nearly match the current top closed leaders. Has OpenAI’s moat dried up, or does the AI leader have something special up its sleeve before the end of the year?

 

By fine-tuning models with carefully curated techniques, TTT improved accuracy sixfold in some cases and set a new state-of-the-art for purely neural approaches, achieving 53% accuracy with an 8-billion-parameter model. When combined with program synthesis methods, the models reached 61.9% accuracy, matching average human performance. The findings suggest that symbolic reasoning isn’t essential for solving complex problems, emphasizing the power of dynamic, computation-focused approaches during inference. → Read the full paper here.

7
submitted 7 months ago by Lugh to c/avs
 

This article - How the Rise of New Digital Workers Will Lead to an Unlimited Age - makes the mainstream case for the future of employment with respect to robotics and AI. By mainstream, I mean that it completely ignores the central question. What happens to human employees when most or all (even future uninvented) work can be done for pennies an hour by AI & robotics employees?

As almost always, he poses the question, and in classic Strawman fashion - pretends to answer it, by answering a different question. Mr Benioff says automation has always created more jobs than it eliminates. But that only answers a different question and ignores the most important one.

Mr. Benioff, CEO of Salesforce and owner of TIME magazine is no different from mainstream economists, or the Silicon Valley elite, in building this world and blindly leading us to it.

One day society is going to have to wake up to the fact we are being duped by these people, and the longer we keep believing them, the more we just get all the angst and chaos, and none of the understanding we need to fashion a new reality.

4
submitted 7 months ago by Lugh to c/futurology
36
submitted 7 months ago by Lugh to c/futurology
[–] Lugh 2 points 9 months ago* (last edited 9 months ago)

I don't mean to diminish people's fears and anxieties because they are extremely real, but I think it's worth considering other outlooks.  For example, look at how quickly the world changed in March 2020 in response to COVID-19. Isn't there something hopeful about that? Doesn't it suggest that the world can adapt to sudden change far more quickly than we expected?

Sometimes I wonder if some people are too apocalyptic in their ideas especially if they come from a product in an apocalyptic Christian background.  if you look at thousands of years of European history isn't the lesson to take away that revolution and change happen all the time, but eventually, progress is what people settle into and things work out in the end.

I realize that is the most hopeful interpretation of events, and perhaps too hopeful, but I'm optimistically natured and that's what I try to stick to.

[–] Lugh 3 points 9 months ago (2 children)

I feel like we are heading to a post-work future where eventually AI and robotics will do most of the work and that will be a good thing. In the meantime, I'm sure there will be a lot of pain and revolution to get to that point.

Even in rich Western countries tens of millions of people rely on driving, delivery and taxi jobs. When people realize they are disappearing forever we'll be one step closer to that future.

[–] Lugh 2 points 9 months ago* (last edited 9 months ago) (1 children)

I don't have sources for the 2024 deflation in China and AI. (I qualified my initial statement "hard to know").

Robotics are a proxy for AI in manufacturing.

I suspect AI is about to give us a type of deflation no economist has ever seen or modeled before. What will happen when AI gives us the expert knowledge of doctors, lawyers, technicians, teachers, engineers, etc etc almost for free?

You can't talk of this scenario in terms of past models, because it's never happened before, but we can clearly see that it's just about to happen to us right ahead.

[–] Lugh 6 points 9 months ago (3 children)

The last 12 months have seen the most sustained period of deflation in China since the late 1990s. It's hard to know how much AI is responsible, but I would guess it is to some extent. It's driving the reduction in prices in the manufacturing of so many things, EVs especially.

Many people assume unemployment will be AI's most destructive economic effect. That may be true, but before it causes a problem, there will be a far more immediate one to deal with - deflation.

Deflation is so destructive because it shrinks businesses' incomes while increasing the size of their debt relative to this income. If there is sustained deflation, then this leads to a spiraling collapse that takes asset prices like the stock market and property values with it. This was the main mechanism that caused most of the damage in the Great Depression.

If AI is on the cusp of giving us lawyers, doctors, and other experts knowledge for practically free, then it follows that there is massive deflation to come. There is already a backlash against AI in some quarters, I would expect it to grow when the deflation problem arrives.

[–] Lugh 37 points 10 months ago (22 children)

If ever there was an industry that could do with some technological overhaul - its housing. 3D Printing threatens to do the job, and seems to have the right tools, but never takes off - will this be the one that does?

At $1,000 per module they offer solutions to homelessness in western countries.

[–] Lugh 3 points 10 months ago

ew data show both have stopped increasing. Is the change permanent? People are planning for this, though it's possible both power sources have a final spurt ahead of them.

This is still a few years ahead of expected schedule so it's hard to tell.

Link to info about China's coal.

[–] Lugh 4 points 10 months ago

When might it integrate Lemmy?

[–] Lugh 2 points 10 months ago* (last edited 10 months ago) (1 children)

I won't be surprised if Chinese astronauts reach the Moon before American ones return to it. Boeing's SLS seems to go from bad to worse, and SpaceX's Starliner is nowhere near ready to completely replace it.

Some people seem to expect SpaceX to work miracles. It has formidable problems to solve before using a Starliner to land astronauts on the Moon. The capability of refuelling Starliner in space, landing on the Moon, refueling there and taking off from it may take to the 2030s to solve.

[–] Lugh 2 points 10 months ago

Last month, Alphabet announced it would invest a fresh $5 billion in its autonomous vehicle unit, which first began as “project chauffeur” at Google in 2009. Jeyachandran told CNBC that the capital will be used mostly for scaling,

That should buy quite a few new robotaxis.

[–] Lugh 45 points 10 months ago

The EU is to change the law to make social media owners and company executives personally liable with fines, or potential jail sentences, for failing to deal with misinformation that promotes violence. That's good, but teaching critical thinking is even more important.

AI is about to make the threat of misinformation orders of magnitude greater. It is now possible to fake images, video, and audio indistinguishable from reality. We need new ways to combat this, and relying on top-down approaches isn't enough. There's another likely consequence - expect lots of social media misinformation telling you how bad critical thinking is. The people who use misinformation don't want smart, informed people who can spot them lying.

[–] Lugh 20 points 10 months ago

capitalist hellscape

It's hilarious seeing Elon Musk taking up the issue of plummeting birth rates, while simultaneously saying people who work for him who won't commit to giving their life to his companies and sleeping in the office are lazy losers.

view more: ‹ prev next ›