- Mehvixhow are we going to deduce/measure/know the initialization and rules for consciousness? do you see any systems as not encodable/simulatable by quantum?
- per halting problem: any system capable of self reference has unprovable (un)truths, the system can not be complete and consistent. consciousness falls under this umbrella
I'll try and ask OG q more clearly: why would the brain, consciousness, be formalizable?
I think there's a yearn view nature as adhering to an underlying model, and a contrary view that consciousness is transcendental, and I lean towards the latter
- you may consider reading I am a strange loop for that, which can do far better justification than myself
if there's surely no algo to solve the halting problem, why would there be maths that describes consciousness?
- There's no reason to believe this to be the case. Godel says otherwise.
- Subsidizing the poor via ads is what we cheer for? bike theif brained understanding
- The gains from MoE is that you can have a large model that's efficient, it lets you decouple #params and computation cost. I don't see how anthropomorphizing MoE <-> brain affords insight deeper than 'less activity means less energy used'. These are totally different systems, IMO this shallow comparison muddies the water and does a disservice to each field of study. There's been loads of research showing there's redundancy in MoE models, ie cerebras has a paper[1] where they selectively prune half the experts with minimal loss across domains -- I'm not sure you could disable half the brain and notice a stupefying difference.
- You're saying it like hardware and software are disjoint. You design hardware with software in mind (and vice versa); you need to if you want performance rivaling nvidia. This codesign, seeing their products are not only usable but actually tailored to maximize resource utilization in real workloads (not driven by w/e benchmarks), is where AMD seems to lack.
Why oversimplify the premise and frame your take as some 'proof'. Just use the term counter-argument/example
- > stuff that's been removed as a "feature" isn't always stuff that nobody wants.
Graphene isn't made to cater to what everyone wants. Face ID and fingerprint unlocking so clearly have no place in a hardened OS. "Google OS-level integration is absent" should not be suprising.
This said, you ought to be able to have BFU security with stock Android and it's embarrassing Google ships stock vulnerable.
- Aye but it is good Apple is safe out of the box. BFU is a low bar, and the shame is on Google.
>Lots more devices are safe BFU than just Apple's
Really? Secure against the exploits and methods these tools 3 letter agencies employ? I hate to cry source, but base Android isn't secure. What devices have similar hardware-level security, or have their Android flavor shipping with these Graphene-OS-level patches?
- >None of that is specialized to run only transformers at this point
isn't this what [etched](https://www.etched.com/) is doing?
- Why do you suppose this is a compute limited problem?
- what's the end game here, have a slew of of finetuned models for these varying edgecases?
- OK but blinding blue LEDs are most common substitutes, because it's the lazy default, and because people do not care. That's the point of the article.
- no, as a kid I was able to go anywhere I desired with a bike or skateboard; I had few boring summer days.
this is much safer/practical in suburban environments. do you really believe public transport, in SF, is as safe for a kid?
e: md fix
- > > People in a functioning society still have to work
.. to pay taxes for social services.
> I'd accept higher taxes ...
how do you pay for these taxes if you have no job/income?
- is/will there be less to do?
in the future, maybe so, but decreasing employment is surely bad during a labor shortage: you do need workers for a functioning, productive economy.
the rise in buying power may look good by the numbers, but doesn't inherently better society -- consumerism doesn't encourage quality goods/services. take AI: it's a lot easier to replace human workers when they've quit, when the positions are already vacant. you don't need to provide on-par performance or quality service(s), just fill the shoes with slop
- >Ticketmaster has a system for transferring tickets, if you want to buy or sell tickets
Sure, and it is terrible.
They can block you from transferring the ticket you bought, and can set a minimum resale price (effectively ensuring you cannot recoup anything)
You should to own what you purchase, simple as.
- >the falling cost of use induces increases in demand enough that resource use is increased, rather than reduced
This is just saying throughput is increased, yes? The time to train, and thus iterate (i.e. dialing in hyperpaprams) will decrease.
- Idea synergy is Ill defined, creating a consistent, high quality dataset of the scale required is not easy at all. What is the data, pairs of syngeristic papers? Or are you embedding each paper and finding the closest in "idea space"?
LLMs are apt for RAG/fuzzy search. Learning a relevant embedding space that has the fidelity where distances have an interpretation of "idea symergy" is not like "semantic meaning" (I.e. RAG embedding space), otherwise you'd simply get similar papers on the same topics as "synergistic"
Maybe I'm confused as to what you mean by idea synergy.
- How would one finetune for "idea synergy"?