To focus your point: we fear that with ideas having easier replication that the noise in the signal will increase.
But this is what people said about the printing press. And the internet. Less effort to spread the memes. Has it been a net negative? Maybe in some respects. Can you stop it? Probably not.
Anywhere the nerds or artists go, if they are good, eventually becomes overrun with such types. Just look at the state of Silicon Valley and y combinator, and gentrification in general…
It will only get harder. From the trenches in tech it’s hard enough to get people who should know better to put in the work to understand the ai output.
For education I can’t imagine. You might have to resort to actual discussion in classrooms instead of homework, with the accompanying need for both more educators and higher quality therein. Which isn’t going to be easy when education is fully under regulatory capture (at least for K-12 in the USA)
K-12 means the public school system in USA & canada. It consists of kindergarten, elementary , middle and high schools, encompassing 12 grades/classes, each taking one year to complete.
Maybe it just requires approaching the problem differently. What about the vulcan study pod things? Have the discussion with the AI itself. Have the AI regularly assess understanding.
So this is an alternative that I actually don't object to at all, but you still need to create mechanisms where you then get at the quality of the mental models formed without the assistance of AI. My (very unpopular) hot take is that some of this will actually usher in a new era of "standardized test"-like things; we will be far more permissive about how/in what ways students/professionals acquire knowledge, and the real task will now shift from that to guaranteeing that there is some competency that was reached with the assurance that it isn't masked by an outside influence.
If you basically assume that everyone can and will cheat at all times, then you can let yourself be far more agnostic about it AS LONG as you are confident in some ability to measure the person's skill without the tools.
The interesting thing is that ai actually helps me learn more effectively than ever before. Each session can become a personalized little seminar that dynamically tests me as it presents ideas and concepts and I attempt to apply and extend them.
I think we need to move past recall as the test itself, and tasks as the gate we judge academic achievement by. These were only ever an approximate measure, which is why we still have things like thesis defenses.
But imagine the final being an interview with an agent that is designed to probe your mastery? From my experience so far it seems we could measure much more directly the educational outcome and also accelerate the learning at the same time. But this assumes those involved want to learn, which widespread cheating suggests is not the case. Maybe the real answer is that we shouldn’t attempt to force curiosity but instead cultivate it where we find it, allowing those who reject it to be redirected until they find something engaging.
I don't miss the point.
Most of the comments that have been posted here seemed to be written by people blinded by efficience and power. None of them took care of my points about the waste of energy and water.
Any technology threating against the natural habitat and its creator is no good.
The same argument has been used against most data center operations from the beginning. You can criticise specific bad actors and bad practices (like Musk and his blatantly illegal gas turbines) without morally condemning an entire field of technology through the lens of religious environmentalism. There is nothing about the technology that prevents data centers from running 100% renewably.
There should be no plan of building new data centers up to that time. I am not against new technology at all, only against indiscriminate use.
It's like cars and motorbikes. People are illuded that they drive good tech, but they are strongly contributing to pollution. This is a topic that should join all people.
Then there's a difficult question to answer: Who should get to use AI and who shouldn't?
The data centers aren't there for no reason, they track demand (and a large helping of hype). It's the same question that arises with cars. A single person driving a car is relatively unproblematic. It's when individual usage adds up that it becomes problematic. And what did decades of car opposition do? A lot less than what happened when a better way to do it (read EVs) came up.
Qwen’s 3.6/3.8 27b actually has some algorithmic advantage when it comes to the kv cache size needed, so it actually needs less memory at equivalent context. My experience using both in vllm supports, with considerably more overhead in context size on these models than Gemma 4 31b, and better performance in most tasks I’ve tried on both models.
Qwen 3.x does have an advantage but it's relatively small (64KB/token vs 80KB/token) - Gemma4 actually has less % of full attention layers, but the largest geometry and has the biggest "fixed" state for it's non-global layers. Muse Glimmer actually has by far the lowest per-token cache usage for the competitive 30B-class dense models - it's at about 13KB/token - very aggressive GQA (32Q/2KV) and also by far the smallest QKV dimensions.
Actually perf (speed) is going to mostly on token output, and here Qwen 3.x historically tends to lose badly as it tends to overthink a lot. I'll be running evals on 3.8 myself this weekend to see how its reasoning levels perform.
this reminded me I gave myself infinite lives in a GNU version of Nibbles (not really the same game but a drastically expanded one) so I could see all the levels.
But this is what people said about the printing press. And the internet. Less effort to spread the memes. Has it been a net negative? Maybe in some respects. Can you stop it? Probably not.
reply