tomohelix

But then you would have issues with how to evaluate the immigrant. You may say degree, work history, other achievements, etc but those are already being used as qualifiers in the current system, in addition to the job requirements. So any attempts at decoupling immigration from employment would be seen as "open border" by many politicians and their bases, i.e. not going to be a reality ever.

If even a law trying to let US-educated PhDs getting an H1B easier (just to work, not GC) get blocked, you know how anti-immigration the current political climate is.

Hilariously enough, a real rocket scientist, e.g. a NASA scientist, has an average salary around 100K. SWE salary is heavily inflated and even the flawed process we have now is better than one where you only look at the absolute amount of compensation. Why, most scientist jobs requiring PhD pay less than 200K.

https://www.glassdoor.com/Salary/NASA-Scientist-Salaries-E73...

https://www.salary.com/research/salary/posting/senior-scient...

I no longer know the current prices for EB2 total cost. But years back I know the cost was >10K+ per employee, accounting for multiple filings, expedited filings, lawyer fee, etc. I doubt it would be below 10K now. Even an NIW can cost 15K easily by going to a law firm and personally haggle the price with them. Can you tell me how do you know EB2 cost less than 10K?

An online paper mill graduate would have a very hard time passing EB2 vetting. Where would they get the paper citations, the impact, the achievements required? Unless they have years of experience in which they made clear they are valuable enough to the field. That is why I said experienced masters. And you think USCIS don't look at what institution the applicant comes from?

Reading into the press release, there is a distinction that probably escapes most native citizens here: they are talking about EB2 visa, not H1B. PERM certification is not the same labor certification done in H1B.

PERM is a much more rigorous and demanding process and it costs a lot more money than anything related to H1B. The reason is because it leads to a green card, not just a work permit. Often, it requires an advance degree and higher qualification than H1B too, PhDs or experienced masters. The money is paid upfront and USCIS then look into it and approves PERM on a case by case basis often taking a year or more. Then when the PERM passes, the applicant can finally get on the green card backlog and wait a few more years, or a decade if you were born in the wrong place...

This is to say the quality of applicants here is very high and Apple actually felt it was worth it to invest tens of thousands of dollars on each of them just for a green card gamble, which the employee can get then quit Apple immediately after and nothing can be done to them because they are now a permanent resident. No such thing as wage depression or abuse at this point because they are for all legal purposes, an equal to any American once they have EB2.

If so, they made a big bet. Vision LLMs were literally made this year. Before that, parsing images to get a coherent response is pretty resource intensive and not really reliable at all. Designing the entire device around image capturing for context seems like a very risky approach so I doubt that was their main reason.

Agree. If they want this to be practical, they have to make it instantly project the image when I raise my hand with palm open towards the device. A gadget that actively make the user feel inconvenient when they want to use it would not be used at all.

I assume the problem is space. They are basically out of space for more features. I still pitch my idea for a smart bracer though. Like the thing in mass effect or the pipboy.

From a consumer standpoint, I do not see the main advantages of this device over a better designed smartwatch or a smart wristband. The main interface, i.e. the projection onto the hand is a very novel and innovative concept but also blurry and cumbersome. A watch would also technically be able to perform all the function shown here. And personally, with how much we are all in need of a smartphone, might as well make something like a smart bracer. That would give all the space needed for the battery and screen and computation and roll everything into one, phone, watch, AI, etc.

Do we have holograms tech yet?

I understand what you are saying so I will address the core of your question instead of just giving you things a child can do.

An LLM right now is still just a facsimile of one of human's cognition. Arguably, without the other senses, it can't experience or understand things the same way we can. This leads to the AI unable to form a comparable sentience to a child, at least, not in the same way and not in a way that would be recognized by most people.

Someone already gave you the sense of time as an example. These senses are things we were born with and critical to our perception of the surroundings. Until the AI is trained on at least the majority of these senses, a child will always have something they can do that the AI can't.

Wasn't there a thing about the mistake of using different tricks and techniques to beat benchmarks but in the end, the product would only be good for getting benchmark scores and nothing can surpass raw computation in general purposes?

I mean the grad assistant is basically a post-doc position isn't it? Considering the stability and perks from a government job, I think that is not too far from market rate.

But go up and those 180k jobs would be easily 300-400K in the industry. I doubt anyone who can do that work would be interested in a measly 180K. Unless it comes with a visa...

You are right, initially I was thinking of a chicken. Then I forgot what branch a chicken is supposed to be in and wrote animal instead. Changed to vertebrate now. Probably closer.

And yeah, this is just my personal view. I didn't say it is what it should be or how it actually is. Nothing sweeping about it. Simply how I feel based on my experience.

The exorcist must be some of the most devoted Christian on the planet. If someone is believing something that much, they can't be considered a fair or reliable observer. Even a staunch atheist when pressured enough and presented with an ambiguous situation (e.g. a weird picture where the shape and shadows just conveniently form a paranormal figure) would start to make false assumptions. Now make it a priest whose entire life is spent on seeing "demons" and I would not be surprised if these guys are going to hallucinate while waking.

A thing stands out to me about the machine learning is their ability to recognize patterns. We all know that human brain is also well adapted to this task. So adapted in fact that it is probably one of the most fundamental, ingrained thing that we have in the lizard part of our brains.

For some reason, to me this kind of subconscious thing hold some special meanings. It feels like the current LLMs are actually going in the right direction to eventually emulate a conscious mind and it shows this by imitating what a mind without consciousness would look like.

Personally, if I have to compare the current LLMs to an organic equivalent, I would say we are at the insect level. It can respond to things, know a few surprisingly complex knowledge (spider webs), but ultimately doesn't "understand" what it is doing and just robotically do what it was programmed to do.

Good thing is an insect to even the dumbest vertebrate is pretty far away in term of mental cognitive ability so AIs aren't going to spontaneously gain awareness anytime soon.

I think because this is literally the same bussing strategy texas was doing and got criticized for, except one is on a bus and the other is a plane. In both cases, the migrants choose their final destination within the US and is provided a free ride to that location with food and water.

One may argue the bus is a much harsher trip than a plane since it takes longer, but that is debatable imo.

There is a balance somewhere though. $60/hr is at the level where some big expectations is normal. You can't demand high pay and low stress work for what is basically a rank and file job.

Has there been any breakthroughs or sign of a breakthrough in term of LLM training efficiency and inference cost? Right now it seems like the capabilities of the models are the focus with more accurate image generation, videos, less hallucinations, etc. But it always at the cost of computing power. Average guy can't afford a $600 GPU to run Stable Diffusion at home at full capacity...

I don't know why a second house or something like that is not being taxed to the ground in California. Like Prop13 was meant to keep seniors from being evicted from their own homes? Sure, that is the primary residence. But what are they doing with the second or third house? That is clearly an economic investment. Get taxed.

Tax those extra houses harder to the point even corporates and landlords feel the heat of profiting off of homes. I don't know why it isn't done. Cali always want raise tax on the middle class but those landowners somehow aren't getting the book thrown at them.

I will take this article with a bag of salt but I think we are very close to solving the human communication problem. Not because the machines are perfect but because the human, our benchmarks, are prone to mistakes and so a "close enough" machine would pass as a human just fine.

Same way it happened before. By being implemented on devices that are not limited to just one port first, prove its superiority, then get adopted on phones where they can't have more than one port.

Laptops with USB 2 ports were still popular after C was introduced. They just get phased out eventually. Your Mac now only has USB-C isn't it?

Better than 30pin for sure, but compared to regular USB, it wasn't that much of a difference to the average end consumers.

And the reason Apple didn't want to go to USB-C is because of money. Selling proprietary, licensed products to hundreds of millions is too lucrative.

The law is about "harmonisation of charging protocols" and currently USB C is just the standard for everything, so it is set as the target for people to "harmonise" to. They don't specifically require the standard to be USB C and if a new tech comes out that is clearly superior, the law mentioned a Commission can be formed to make a judgement on switching.

Page 17 of the law document.

> The Commission shall continuously assess market developments, market fragmentation and technical progress with the view to identify categories or classes of radio equipment capable of being recharged via wired charging for which the inclusion in Part I of Annex Ia would lead to significant consumer convenience and reduction of environmental waste. The Commission shall report such assessment to the European Parliament and to the Council, for the first time by [3 years after the date of entry-into-force of this Directive] and every 5 years thereafter, and adopt delegated acts pursuant to Article 3, paragraph 4, first subparagraph, point a) accordingly

I don't think the first iterations of lightning was far better than USB 2. It has some features but overall was about the same. Then USB C came out and apple was left stagnant for more than 10 years with zero innovation.

Also, the new law has nothing against making a better port than C so I think you are mistaken.

Apple? I hate that company for the anti-competitive practices. They literally tried making everything proprietary. It took the entire EU and a targeted law to make them adopt the USB-C.

Do those chemical engineers make the same amount of money as a software dev with the same amount of experience? A fresh ChemE grad would be lucky to land a 120K job as a process engineer. Very lucky. An entry software job is 150K. Does this mean the a CS grad is on average more impactful to the economy than a chemical engineer?

Not all industries can pay as much as software. A PhD with an extra 5-10 years of industrial experience may get a 200K salary in biotech, often less. A CS bachelor with 1-2 years of experience would get that 200K. Pay doesn't always equal the value of your contribution, especially if you compare them across different fields.

Because software isn't the only industry critical to an economy, others are too. You can argue that pay=value to the economy but in practice, you can't equalize value and pay across all industries. Here is an extreme example: no matter how good a high school teacher is at teaching and innovative classroom management, no school would pay them 300K a year to teach because the average salary for this role is 60k. A more realistic example is a biotech PhD scientist with years of experience and proven impact in pharmaceutical industry. You would be hard pressed to find a company willing to pay this scientist 300K a year (avg salary for principal reseach scientist in SF is 200K).

Value and pay aren't the same across industries. Someone can still contribute the same value but the field they are in simply doesn't have the economic profitability required to pay them the same way software does. In fact, software is the anomaly in compensation. Making a policy that benefits only one type of industry over everything else is simply bad policy.

Honestly, I would rather be with someone who shower a bit more than they should than someone who doesn't shower enough.

Fact is, smelly people don't think they smell until it is pointed out to them. Sometime these people are from the same background had never been outside of their own group and so they never know it for their entire lives. I have seen plenty of time when just one person can stink up the entire room and everyone is too polite or just too afraid to call it out.

Those meetings were hell.

Less side effects. But also much harder to implement. Especially when the people with the power to change things are incentivized not to change it.

So let just settle for the next best thing while we ride this clown show for as long as it can last. Because let be honest, the US gov is still doing an OK job compared to some truly dysfunctional ones out there. Not many Americans are dying to emigrate yet.

Please give me an actual example, not a 17 years old video of a theoretical assembler that is still not a reality yet. From what I know, all of this is purely science fiction and worth nothing.

Although to be honest, there are indeed molecular assemblers already, except they are biological...

From what I read in your post, I must ask: what is your credential in this field and are you discussing this with good faith? It seems you have a very basic understanding of the mechanisms going on at the molecular level if you think enzyme is just about things randomly bumping into each other. Also, your description of kinesin mechanism is completely wrong. Kinesin doesn't move backward without ATP. It doesn't move at all without ATP. And it moves in one direction only unless its substrate was modified or external forces applied to it. Calling it random walk is ignorant at best and dishonest at worst.

>This is not what happens. If you actually watched a molecular dynamics simulation of a typical enzymatic reaction, and slowed it down just enough that you could see molecules moving around, it would take YEARS

There are so many things wrong with your claim. First, the same binding based on Brownian motion would happen regardless of a biological or synthetic nano assembler. All of them requires the substrates to reach them first before they can proceed to the next step and this depends on the kinetics of the reaction. Thermodynamic doesn't care if you are a polymerase or a tube of graphene. DeltaG is all that matters. But an enzyme can have specific interfacial and hydrophobic interactions built into its structure that promote binding and increase the distance it can "grab" another molecule.

An enzyme is a catalyst that lower the activation barrier. A regular reaction, which may take a day to proceed based on random chance a molecule gets a quantum fluctuation in its energy state to overcome the barrier, would get over it in an instant with an enzyme. It positions and moves the atoms around such that the reaction becomes much more favorable. That is the whole point of catalysts. An enzyme conformational changes are deliberate and extremely precise. I have said this multiple times but you keep ignoring them. Please look up the word if you don't know what conformational mean. These happens extremely fast, at femtosecond timescale, at the length scale of a few angstroms.

https://pubs.rsc.org/en/content/articlehtml/2018/cc/c8cc0242...

I didn't go too much into this because we in the catalysis field work with turnover events, which is the operation definition I gave you above. Going down to the elementary reactions is often too fast and difficult to work with. You literally can't do MD at the precision required to model some of the enzyme conformational changes.

And what do you mean by slow them down and weeks and months? If I slow any video down they will take longer to play. What matter is how long does it take in realtime. I gave you a Mhz speed enzyme. That is its real speed. It literally processes a million molecules per second by ripping out a single atom on each molecule. What do you mean by weeks and months? How are you alive if reactions take weeks to run in your body? In fact, please show me a Ghz synthetic assembler that can do the same to a molecule. I am ignorant of any but it has been some time since I was last involved with this type of research.

Honestly, I tried to give you some benefits of the doubt. But after typing all this out I afraid you and I are not talking on the same level here. So here is a simple ask: give me a real, proper, actually proven, example of a Ghz synthetic nanoassembler, a nanomachine that actually take a molecule and move it somewhere, the same way an enzyme works. No "theoretical videos" or baseless claims.

This thread won't end well... People on both sides has pretty strong opinions about the California culture. And if the problems don't bother them that much, they defend it. But the same problems might be dealbreakers to others, and those will heavily criticize it. And I doubt there is an objective truth here since it is about people's perception of a human culture, both are literally definition of subjective.

>directed motion, often carried out in vacuum by parts highly constrained to follow rigid paths. Like a factory

That sounds exactly like biology to me. That is how any enzyme operate. Why do you think it isn't? How else do they literally pluck an electron from a molecule and move it to another molecule at precise atomic positions? How else does biology reliably produce polymers with polydispersity of 1 if it isn't controlled by highly directed parts?

Regarding the speed, I would like some clarification. In my field, when I mention an "operation", it means the machine has to work on a physical object, process it, and output a product different from the output. These are often molecules like drugs or some chemicals. So when you mentioned Ghz operation speed, are you talking about a current nanotech that can actually produce billions of products per second? Or did you mean it just vibrate or have some repeated motion? I am interested to know that.

Here is an enzyme with 2Mhz operation speed whose parts are just exactly one atom: https://en.wikipedia.org/wiki/Catalase

And for a more apple to apple comparison, the fastest moving nanomachine I know moves at around 100nm/minute

https://www.advancedsciencenews.com/new-dna-origami-motor-br...

Compared to a kinesin, which isn't the fastest protein, it is over 800nm/*second*

http://book.bionumbers.org/how-fast-do-molecular-motors-move...

I disagree. Things bump into each other randomly but that is it. An enzyme conformational changes are anything but random. And the thing can work at the speed of a thousand molecules per second. That is not slow and I guarantee you an engineered nanomachine is nowhere near that speed.

Get down to the molecular level and biology is at the machine phase.