I had to return iSpring RO system bought through their official portal. It came as a shock to me that I had to pay for shipment that was 70% of the contents value. IMO that's the way to discourage returns.
When it can replace a white collar remote worker without humans team realizing they work with the machine. And no, Agents are not like this whatever initial prompt and set of skills you give them. They still won't progress, learn and apply that knowledge like a human.
> A dumber, less social world, is far less likely to be a successful world, even if the tools available are unprecedented.
En masse such worlds had successes in the past - renaissance, industrial revolution.
It's something else what I can't describe but it's the zeitgeist that was different when world recorded new successes. Look at CS revolution that led to PC and web of nineties and noughties, they didn't think about the result product , or how to steer thousand engineers to build something - amazing things were born in a very small teams, many times authored by a single person, who was deeply invested into the field and knew what he was doing.
[1] is not a valid proof LeCun was wrong, LLMs still can't do spacial reasoning when it can't be derived from the training data. He didn't argue that GPT 5000 won't be able to describe something with words.
This really doesn't match my experience. I can ask an LLM to modify engineering plans using vague natural language prompts and it will find the right place in the plan from the description and then make appropriate modifications, which necessarily requires doing spacial reasoning.
Or it’s just taking common examples from training and applying those copied heuristics to your problem? Doesn’t mean it’s actually reasoning about the space and how to solve the problem. It’s the equivalent of a student writing an answer they saw somewhere else without understanding “why”.
Why does CoT significantly improve their performance? Most of what they are applying they learned in post-training by solving similar problems themselves. This isn't about regurgitating pre-trained knowledged.
Advancements in Math and coding are because RLVR at massive scale is so cheap.
I have not seen the research on CoT and how it impacts spatial reasoning and outcomes, so I can’t comment on that part. I do see continuously that in almost every example of non-trivial image generation and 3d modeling, there are quirks that point to the fact that the model does not understand relationships within the space based on physics.
CoT seems to work really well for text based generation but once you’re past that and into physics models and detailed relationship mapping, it may work better but it’s not enough to make me believe it’s “reasoning” in a way that humans do.
Looking at the reasoning traces it sure seems like it's reasoning. It internally debates which of the possibly matching parts of the input are the one described by me in the prompt and picks the right one based on sound reasoning.
It's not just text that appears at first blush to resemble reasoning, it's actual sound reasoning. And it can chain it for hours at a time without breaking down.
I think the point here is that LeCun was arguing that training on pure text would not grant spatial understanding. I believe most models are trained on spatial data as well, so you are both right.
Is that what he meant? He works on models with an explicitly spatial internal representation, whereas I was using a standard LLM that edited the provided plan by using a bajillion python calls to inspect small regions of the image at a time and then generate edits.
I've seen recent AIs make detailed and technically impressive 3D models. You might argue "they're not doing spatial reasoning, they're making measurements with code and doing math to configure relative positions". Fine, but at a certain point that becomes functionally indistinguishable from spatial reasoning.
When put to the test in real-world environment, the capabilities don't look as impressive as benchmarks and synthetic tests might indicate. So doubts about actual spatial reasoning capabilities remain.
Aaah, the old benchmarks maxxing argument, having precise and clear definition of what "spatial reasoning" is, what, and most importantly WHY, the benchmarks of choice are would settle this debate, otherweise let's not delve into it.
This advice helped a lot of students I interacted with - hypes come and go, observe them but think about yourself in 10 years, invest into your skills, even if that means writing web pages manually for the sake of learning. Think about all those who invested into UML code generation, Java EE frameworks instead of core skills.
Most houses are old and many in shit condition, you'd want at least replace shingles before installing solar.
Everything becomes harder and expensive - no more standard gutter cleaning fee (200-300 bucks per average roof). For damp subclimates no more moss treatment possible.
For many houses you'd need structural engineer assessment to make sure roof can add such a load.
Many vents and outlets need to be relocated to cover large and nice area with the panels.
Dealing with contractors on standard things is such a pain already in US - expensive, many unskilled, trying to get more money out of your pocket. When they see someting expensive or non-standard their attitude changes even more.
AI can make everyone miserable. Solution? Partner with all major AI companies using foundation (tax free btw) money. It's like saying Hitler is bad, let's split Europe in half so we can profit as well... ah well, that actually happened.
User experience is not always correlated with metrics targeted by engineers.
I visit a building sometimes with destination dispatch elevators, and with low exposure across multiple years I'm confused every time the elevator I need to take is open behind me.
I once used the paternoster lift in Germany, beats any elevator algorithm efficiency because it's a non-stop "conveyor" with no doors. I still get nightmares about that elevator.
reply