The US government can't legally pay the salaries required to find AI experts right now (just look at the salaries on the open positions). So all this work will end up being done by contractors who _can_ pay those salaries while the contractors skim off the top.
The US really needs a special schedule for software developers if they want technical expertise in house (and they really should have technical expertise in house)
Recoding America[1] is an excellent book about this problem, but it extends a lot further than just contractors charging inflated rates. The government itself has a culture that is so averse to trying new things and compliance with existing policy that it's actually dangerous to the careers of civil servants to do anything innovative, because that is one of the only things they can actually be fired for.
The solution she proposes effectively would be an expansion of the US Digital Service (which the author spearheaded) into something that approximates the British civil service, which is actually highly functional and effective.
If you actually care about this problem and have a spare Audible credit it's worth a listen.
I was recruited by, and damn near did a tour with USDS.
I ended up passing on it.
They wanted an "entrepreneur in residence" to help drive change, which sounded great to me.
Then they did the whole "we need you to keep in a cup" thing.
I wouldn't have had any problem passing the test, but I told the recruiter "this is where the change needs to start".
There's no way I can attract and retain good technical talent if we're still ruling out candidates because of smoking pot.
Plus, I find the whole "pee in a cup" thing demeaning, intrusive and offensive.
The recruiter and I were both very frustrated by the situation. I wasn't the first candidate they'd lost over the issue.
I don't know if it's still like that, but the feds seriously need to get their act together in hiring process before we can improve the quality of government systems.
Current USDSer here posting personally. I hear this and you aren't alone. It's not a decision specifically about USDS, more about the entire White House staff and federal law.
I mean no offense, but this is sort of, with regard to your principles at least, cutting off one's nose to spite one's face? I get that incremental reform is frustrating, but here you had a chance to enter an org you know needed changes, and you blew it because you refused to take a basic drug test that you openly state you could've passed easily.
Like, it's bullshit that they asked, it is an invasion of privacy, it shouldn't be a reason to forgo employment of a candidate they otherwise liked, but at the same time, it wasn't an issue of like, you wouldn't get hired because of your race or gender, it was an issue of like, you'd have to abstain drugs for a while. And then, once you're part of that organization, you can help push for the exact changes you want to see in it.
It's the same reason I get frustrated with my generation not wanting to run for office: we spend all our time complaining about how awful the government is, but precious few are putting themselves out there as the instruments of that change. I don't know if it'll work out or not but I know damn well it won't if we refuse to even try.
Almost all bad business practices were ended by labor refusing to comply with them, not complying to get the job and then somehow changing it within. I think that on top of not taking a massive pay cut, not submitting to a test that can flag you as an opiate addict to the government for life if you ate a poppyseed bagel in the morning is reasonable.
I get that perspective, and I did think about it a lot.
Understand that the position was mostly charity work - the pay was negligible for a six month contract. I was ok with that part - I was motivated to help, not by compensation.
The intrusive testing, coupled with bad compensation is a root cause of the problem IMHO.
I think organizations only really change when they feel pain. In this case, my hope is that the pain of repeatedly losing good candidates is enough to drive policy change (at the congressional or executive level).
I don't think it helps to try to "change from the inside" in this case. If I (and other candidates) just roll with it, there's no incentive for change.
You are completely right. Once inside, your power to affect change is nearly zero unless you are a very high ranking person, and even then it's pretty near impossible.
The change will come when enough people say "we can't hire the best people!" Until then, the only argument is "it's an invasion of privacy and outdated" but if it's not broken (i.e. not causing any problems) they won't fix it. All the candidates power is at negotiation of the offer time. Once you accept, it's kind of over.
I'll add, from their perspective, it may also have security risk associated with it. If you are applying for a federal job, there may be a chance that you deal with sensitive information. Just like I may think it's bullshit if they ask how much debt I'm carrying, I understand it from the standpoint that it may impact my security risk.
It is as well in most of europe. And its also illegal to make your employees take a drug test in most of europe unless you work for the police or with heavy machinery
I actually considered applying for the US Digital Services, I was okay with the salary decrease as long as I could continue to work fully remote, but they require you to work in DC. Telling ppl to take a huge pay cut AND to have to show up to the office is a huge ask.
> As of April 2023, we began formally hiring remote as well as DC (local) based employees who report in-person to our D.C. office or the agency they are partnering with.
I'm also fine with the salary decrease but at least with USDS they have you work a "tour" (like a tour of duty, I guess) which lasts between 3 months and 4 years. First, that's a little ambiguous; who decides the duration of the tour? It also feels like it would be off-putting to get to the end of your fourth year and it's your best-performing year so far, but you're just Logan's Run'ed out of there. Even more so because neither Congress nor the Supreme Court, nor the overwhelming majority of state Congresses, have term limits.
> With tours of service lasting no more than four years, the U.S. Digital Service brings fresh perspectives on technology and delivery to the government.
Feels like " 'fresh perspectives' for thee but not for me".
Representatives have elections to determine how long they can serve; career civil servants have strong protections against firing but the downside is a challenging hiring process that can be very slow.
USDS's willingness to use term hires makes hiring folks in easier, in addition to ensuring that the office doesn't become stagnant.
It was explained to me that they have to do the "tour" model (which is a minimum of 1 year, up to 4 years) because it is the only way they can bypass the usual lengthy government hiring process.
so what happens at the end of the tour? you get interviewed again? or are you barred from rejoining unless you do the lengthy government hiring process?
> that approximates the British civil service, which is actually highly functional and effective
Not when it procuring large systems, IT or otherwise. As routinely discussed and complained about by the National Audit Office (NAO) [0]. The civil service relies massively on external consultants and contractors to achieve anything.
Sadly our track record on IT is disgraceful, yes. Billion pound IT
black holes are strewn across health, education, defence.... It's a
legacy of cronyism and corruption from the incumbent Tory
"government". Consider Sunak's connections. OTOH our Civil Service
remains a beacon of competence in a sea of insane policies,
reshuffles, resignations, criminal scandals and total absenteeism.
It runs on "values", plus a bit of spit and prayer. There's no money.
Running something equivalent to the International Atomic Energy Agency
but for AI will need funds and support of the same order.
Which parts of the UK Civil Service would you say are more competent / effective in your personal experience? From the outside (US) perspective it's hard to see it as anything except the same-old bureaucracy and crazy rules that make it impossible to have any positive impact in America. (For example, the UK rules for getting a passport basically make it impossible to find somebody to countersign your application unless you happen to know a lot of lawyers / doctors professionally)
That's a good question, because I'm speaking now as a regular citizen,
civilian, and discounting any "professional" contact I have with
government. A lot of it is feel, and also by comparison to how _bad_
it once was, and the kind of thing I've experienced in mainland
Europe.
In Britain the cs are basically providing long-term continuity
regardless of how chaotic and volatile party politics becomes. What we
see is that most of the machinery of state keeps working, despite the
circus in the Commons. Despite Covid. Despite Brexit. So, without
subscribing to an anarchic take, we somehow concede that somebody,
somewhere, must be doing some work :)
Most people's contact is through online, if you can fire up even the
most basic browser almost every interaction is clear, fast and smooth.
When you start dealing at a higher level (business and contracting)
then you see behind the Wizard of Oz's curtain a bit. Things go more
smoothly if you have the right documents and history. Isn't that the
same in every country though?
So, it serves regular citizens very well, but less considerately as
your needs are more complex. Effective maybe, but I didn't say "fair"
:) Certainly there must be many people find it harder, and I'm
probably showing my privilege.
Yet somehow the same institutions produced gov.uk, which is superb compared to just about every other country’s counterpart. I’ve no idea how you did it, just wanted my appreciation on the record :)
>it's actually dangerous to the careers of civil servants to do anything innovative
As a general rule, this is probably true. But there are certainly roles where people are encourage to innovate. Remember, the government is huuuge. If you get a job in a legitimate research lab, you are not only encouraged, but expected to innovate. I'd be careful we're not conflating something like an operations role with that of a research role. (Although, to your point, there is some value in ops people innovating, too).
As someone who has worked at such research labs I disagree. Innovation is encouraged, sure, but the research funding model, agencies, and most research labs create environments that push work on small iterative efforts and system gaming.
We as a society have stopped making sure our folks have overall stability in their jobs and as such, maintaining stable income for most is a priority. We don't accept failure, we don't proportionately reward risk taking (and instead, promote risk passing/mitigating), and were increasingly forgetting that regular failure is part of life. As such, even "innovative" labs are increasingly becoming less innovative, IMHO.
Where you see innovation happening is where there's typically piles of capital that provide people the stability and environments to take risks and fail, combined of course with luck or foresight in choosing the correct paths forward. That or in places where people sacrifice aspects of their own life to move something forward through some passion, obsession, etc. which is not a market of people I want to compete with.
Yes I meant people in government who are responsible for spending money and overseeing programs. NASA and NIH scientists and DARPA PMs are amazing innovators that should be celebrated. All of these people are responsible for an incredibly small piece of government spending.
> she proposes effectively would be an expansion of the US Digital Service (which the author spearheaded) into something that approximates the British civil service
Which is pretty much what the proposed regulations on AI will do as well whist destroying small company innovation and treating consumer AI like software and media piracy.
I guess the president felt bad about banning GPU sales to China so he decided to gimp America too.
We spend plenty of money on ideas. We need to pay for results. We should stop pretending government is ever going to be good at buying/building tech and change procurement to pay for innovative results. How many $50 or $100M government tech projects have to fail before we let them post an RFP that basically says do this thing in this amount of time for this amount of money with payment after we get the “thing” in time. I know an entire industry of innovative people (tech startups) who would find a way to get it done.
That approach works for buying a WordPress deployment or a Gmail license. It doesn’t for anything where you need customization, serving needs or audiences which the private sector doesn’t prioritize (the VA can’t blow off accessibility or tell veterans that they need a fiber connected-M2 to have decent performance), or provide income to a startup which needs to explore a somewhat novel problem.
The real problem here is one you see in many organizations: misaligned incentives. Government has some unique challenges but most of what you see are common in the private sector, too: some people talk like the public sector is consistently super high quality on lean budgets but for every Chrome there are a dozen terrible enterprise apps which also fully count as private sector products. The pathologies are mostly the same: internal politics, building things you won’t personally use, not wanting to revisit past decisions, or otherwise having your paycheck come from things other than what’s best for your users.
Startups aren’t magic pixie dust for avoiding those problems: most fail, and the winners often hit the same problems from a different direction - making users happy until Google buys them and cancels the good parts, having to start aggressively monetizing at the expense of the user experience, etc.
Going through the procurement process itself results in massive delays (have you tried bidding on a federal government contract?), a race to the bottom (since the government more or less has to accept the lowest bid), and massive overspecification by policy writers (mostly to try to salvage some quality from these RFPs).
Echoing that Recoding America is a great book to read even if you have nothing to do with the federal government. You might see some similarities in how large companies operate.
> we let them post an RFP that basically says do this thing in this amount of time for this amount of money with payment after we get the “thing” in time.
These exist; they are called "firm fixed price" contracts. You still have to write the contract correctly to be outcomes focused, and that can be hard when you don't know until building/testing if you are building the right thing.[1]
Trouble there is whenever spending is promised for new ideas, it becomes campaign fodder for someone to stop that spending because "taxes are too high" or whatever line gets conservative voters foaming at the mouth.
It's an uphill battle to get new funding, and talking points move faster than budget approvals and program results.
I'm not sure if entirely or "too" conservative or progressive organizations or governments succeed.
Bureaucracies who can't change when needed, fail if they are stuck on constantly "conserving" at all costs, or "changing" at all costs.
There is some neat research out there that I will try to dig up on how corporate organizations actually benefit from having a balance between the "go" and the "whoa" in an organization, whether it is conservative or progressive views on something. This is not to say that those primarily obsessed with converving (keeping the world how it is, for the benefit of who is already benefitting only) may not be progressive and expand, nor does it mean that progressives aren't capable of restraint.
Making it about sides, is really, really, oversimplifying it though. It's one of the thing I think that is really polarizing when people who focus on differences more find more of them, but if you ask them, they ultimately have a different way of having the same thing in common. Maybe this is unique to the US, but it has always struck with me how people align completely binary around right or left when so many folks if they took the time to learn where they stand would probably end up closer to the centre on one side.
Fanaticsm at the extremes of any interpreted ideology is a bigger issue. Those folks can neither change their mind, nor change the topic, nor openly entertain a viewpoint that isn't theirs, but try to be open minded. It's my hope that societal progress can make inroads here because poeple do have more in common than they think.
Bureaucrats are so risk averse to such an extent that the culture can have an allergy to trying to do things different or better too quickly (including innovation).
Sincerely asking, does your question intend to pre-suppose that everyone must have and express a unique perspective and no one can independently have anything in common? Or some other faux pas such a comment would run a-foul of?
When I saw original post, I had this comment. There was kinship with the parent post. Sometimes echoing your support with another comment is a valid comment too.
If you may want more novelty and context and details:
My professional experience includes working across multiple industries of bureaucrats, and that was my visceral reaction to the piece. Change usually means studying change, analysis, design and ultimately decision fatigue by committee.
There's many hard working and great bureaucrats. My statement tends to apply to the leadership and management layers. In terms of keepign things the same or very slow and tiny changes incrementaly, there's a lot of neat reading around institutionalization as it relates to not only government but bureaucracy here.
An exciting part is how the pandemic challenged and made it OK for bureaucracies to try things and to roll out programs that weren't fully thought out or known and update them regularly. In environments many people could not handle navigating.
Most of all I'm flattered by the downvotes.. happy to take those for the parent too. Perhaps I phrased it in a way that might have triggered some bureaucrats since they exist in enterprise, government, education, healthcare and more.
Or maybe it bothered the dopamine seekers in only novel and unique comments have worth.
Some of the jobs go up to $180k/yr. That's not too bad considering it doesn't look like it requires DC location. But others like the grad assistant for $22/hr is laughable.
Still, I agree there's a problem with salaries not being competitive. I've interned at DOE labs before and lots of people don't stay for long.
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...
I think you'd be surprised. Plenty of people are interested in going into public service and especially policy related work. The high industry salaries can actually be a positive here, they allow workers to be FIRE'd in a decade or less. At that point they can switch to a lower stress/higher impact gov job and the salary being lower isn't a huge problem.
Statistically, this doesn't really happen though. The % of people who make enough to adopt a realistic FIRE strategy are not going into govt service in large numbers, not by a long shot. Even the more traditional retiring early crowd isn't going into public service
The govt has record vacancy at the local, state, and federal levels for jobs[0][1][2]
Even taking steps to lower barriers to entry, they can't fill them.
I think the key here is that the job has to be intriguing enough. Working on shaping US AI policy is something that many people would sacrifice quite a bit to do. Pushing paper not so much. Most fed jobs are on the paper pushing side of things, which also makes the actually interesting jobs harder to fill due to the reputation for bureaucracy. Probably why they made this fancy recruiting site.
You won't be able to FIRE on $180k for 10 years (or less), not even close.
The thing about FIRE is you actually need way more money than regular retirement, because you have to plan for a 40+year retirement instead of a 15-20 year one. On top of that you need a higher margin to account for uncertainty.
HDThoreaun is saying that it's possible to work in the private sector for a decade and save/invest enough to FIRE, especially if in your "retirement" you're also pulling in $180k/y and getting government benefits.
There's "barista FIRE" (save enough and then go work as a barista, typically at starbucks because they had unusually good benefits, but I don't know if all that has changed after the unionization), but honestly working for the government seems like it could be even lower-stress than being a barista.
The sentiment today is the idea of the post doctoral position is also exploitation. Paying PhDs to do cutting edge work for $50k a year is a joke. In n out managers probably get more than that now.
Almost all those 300K+ industry jobs are willing to sponsor visas.
Most of these jobs on the other hand seem to be US citizens only. Some require a security clearance which a lot of AI experts would probably rather NOT have.
> 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.
At a DOE lab in my second year of PhD I was getting ~$50/hr. When I've interned at other places I've gotten to about $60/hr. That's still below the typical rate of $80-$100/hr for a new grad (we're just talking base salary, no comp or bonuses). Definitely not market rate.
For the $180k, that's a bit tougher. Since this is all cash. I'm not aware of anyone who pays $300k all cash to newly minted PhDs except for Netflix. Otherwise that number is competitive with typical base salaries. But like you say, with equity that number doubles and so isn't competitive that way. There are benefits to a gov job (way more relaxed) but you're turning down a lot of money, that's for sure. And of course, the salaries are only competitive for newly minted PhDS, not for anyone that's been in the field post PhD for 5+ years.
$180k/yr isn't bad in the context of Sr SWE in medium-cost-of-living areas. But in the context of ML/AI, it's pretty rotten no matter where. Hard to get the right people if that's the top.
"Up to" $180k for expertise in a field that is as technical and in as much demand as AI is low. For experts in AI, I wouldn't be surprised if their total comp in larger tech companies is around $500k-$1M.
You can expect to find some experts with a stronger sense of civic duty willing to take a pay cut to work for the government, but when you're talking about multiple multiples of your salary, you're going to be left with people who are not top talent or those who are already so wealthy from other jobs that they're willing to work for much less.
Not to mention, that 180K is either barely more (on the lower end) or less than a senior web developer makes at a good sized company. Thats just how out of sync government salaries are
Usually too, the top end of the salary is never going to be offered for a govt job, because they need to be able to give you raises at your level before your next promotion. There's lots of hoops that you end up going through with gov jobs that don't exist in private industry, like pay band caps and promotion schedules that are most often seniority based rather than merit based.
It is stable, and the retirement benefits are really great. The healthcare used to be spectacular but for newer hires the healthcare benefits are more in line with what you get at a private company with good benefits that isn't a FAANG. IE, you'll still have to contribute, the government doesn't pay 100% cost of insurance anymore.
source: I used to work for the government, and know many government workers via my past work, friends, and family at the state and federal level
Having previously been an AI scientist hiring manager, that's around the starting salary for an AI PhD fresh graduate in the USA, which does not include equity.
Where exactly are you seeing that $180K remote job? When I specify 'remote' on that site only 4 jobs come up. One of them is data scientist, another is IT Specialist and the other two seem administrative. The IT Specialist job is listed at $99K while the other three are around $70K/year.
Some say wage suppression is one way the much better paid bureaucratic class can keep their well educated potential replacements where they are, far away.
This is most acutely felt in acquisitions where those responsible for deciding what to build and buy have very limited understanding of software development. The failed rollout of healthcare.gov is the billboard for this problem.
There's an effort to train civilian acquisition professionals on how to buy and build better software, but they are pretty quickly hired by contractors or the commercial sector since their new skillset fetches a larger salary than what the government can pay (even with promotion).
I agree there needs to be a special schedule. I do think an overhaul of the civilian service is worth exploring.
second hand knowledge in California is - the State is large and wealthy enough to have some echoes of the actual Federal govt.. and repeated words from direct software project participants was .. that mid-level bureaucrats from the "controlling" contract originator, would spuriously and without warning change the requirements.. sometimes so much so that it would invalidate large'ish portions of completed work.. and this happened all the way into the final month of a supposedly serious deadline, with failure to perform clauses and all that, at stake.
A first-hand voice from California State software projects said - that the required prime contractor had 10+ person teams and $1m+ budget, and failed to build useful results. The final result from the project(s) more than once, was "oh oops, doesn't fit with -other framework- cannot be deployed". In other words, literally making things that are abandoned on Day 1 of deployment, for large money, with plenty of apparent pressure on "workers" to perform. etc..
This is probably an unpopular opinion in tech circles, but perhaps the goal of top tech talent shouldn't always be centered on amassing the most money?
I get that the pay scales don't hold a candle to private industry, but they aren't pauper wages either. They're in the top 10% of the US as a whole. I suspect if that isn't sufficient, it might indicate a difference in value systems.
We've abstracted so much in life as a transactional situation through which our system of exchange, money, holds power to do so much. It shouldn't be surprising that the some of most useful resources to accomplish tasks in life or even necessary to survive are so sought after.
If we adjust the incentives of life a bit so people aren't constantly looking for another buck to make sure they can actually retire or pay their mortgage off, assure they have access to healthcare when they inevitably get sick, and so on we might find a lot more people would invest their time towards other contributions in society rather than those that they find that compensate them the most.
I agree that the transactional mindset is part of the problem. The irony is that the research seems to imply that an awful lot of our happiness can be improved from taking a different mindset.
As someone who used to work in government contracting, this comment is so freaking true.
The whole thing about contractors is so spot on. The only slight difference I would add is that the contractors will NOT hire the best talent, they will hire whoever they can to fill the position, for the cheapest price, so they can claim to have fulfilled their RFP.
I haven't done much, but anything I ever deal with that involves the government and contractors is the same. The contractor has an astronomical margin and, in addition to not hiring the best talent, they don't care about their employees and have a high churn rate.
These are also the same ones that are super eager to have AI replace their (unappreciated) employees. They don't care about quality at all and as long as they fulfill the contract with something barely working that's all that matters to them.
I predict things are going to worse and worse when systems don't work. Having failures in critical institutions that are life altering for the people that are unlucky enough to be the outliers is going to become the norm IMO.
> and they really should have technical expertise in house
Why? Moreover.. how would any of this expertise actually get applied top down in a federal system? How would any department making purchases or technology decisions actually benefit from such an apparatus? Wouldn't it be more sensible for them to manage those purchasing and operational decisions the way they typically do? Having the expertise being more local to the actual deployments?
There's this "federal czar" ideology that comes from the 1950s, but I think the spread of technology and social development has reached a peak where this centralized expertise model actually holds us back. If we are /actually/ at a point where something like "ai.gov" should credibly exist, then we're definitely at a point where centralized bureaucracy isn't an adequate approach to manage outcomes.
Working for the government should be in the same ballpark but typically work at the government pays a less as there is to be the perceived benefit of social good imbedded in your salary as well as there are power considerations (in this case impacting AI policy). That and your job is generally considered to be safer than working in private interest. The incentives aren't the same.
For example judges take on their roll even though they would be paid significantly more in the private sector as lawyers.
I'm not sure about the USA, but government jobs in Australia also are very stable with strong contracts and benefits (in general, every department is different).
Yep, the US government gives contracts for a lot of stuff like cybersecurity etc. because the GS pay scale can't compete with the private industry standard.
Not to mention that you will have your livelihood at risk because the Congress and President can’t agree on a budget. I’ve considered working for the government. But I’m not willing to put my paycheck in the hands of incompetent elected representatives.
I wonder if there's some way for the government to partner with one of the major tech co's (e.g. microsoft/google/apple) similar to how the military seems to partner with private co's (e.g. lockheed martin).
They can but that has several widespread problems: the biggest is that you need skilled people in the government who can review bids and oversee work – otherwise you get a $10M contact form because someone at BigCo1 wanted to play buzzword bingo and keep their buddies employed, and the “overseer” from BigCo2 wasn’t going to saying the team, often former coworkers, wasn’t performing well because that’d invite honest grading on other contracts where the relations are reversed.
Beyond that, you get various efficiency issues: contractor turnover tends to be higher, which is terrible when institutional memory matters or when you’re relying on their best people to keep a larger group productive, and there are often various contractual or policy issues making work less efficient. One challenge is that managing contracts is a lot of work, so there’s a temptation to go with the big companies who offer many services but that often means that they’re only good at one area and you get mediocre results in the other ones which were lumped into the same contract - often banking on the difficulty of taking corrective action meaning that people will accept low performance in non-core areas to avoid jeopardizing a core project by the same company. None of that’s intractable but it adds up and fixing it is hard.
So far the support of Apple, Google, Microsoft, Facebook and Cisco/Akami have given the US government a business, intelligence, and social engineering platform the likes the world has never seen before, and still people today wildly undervalue its power in the world today.
They have effectively deplatformed Europe, South America, the Middle East and Asia, put anyone but China within second reach from a mass data collection perspective, and expanded the economic hegemony of the US more deeply and completely than the military ever could.
The US is now on the verge of the second great breakthrough of mass automation of that data, intelligence and action. I am honestly shocked that this third stage of American Dominance has not be countered by any other nation save China, and even they're only defensively. America's global brain drain might have been the single most effective strategy that the planet has seen so far...
The only question is if the rest of the world is going to wake up and counter this.
> America's global brain drain might have been the single most effective strategy that the planet has seen so far...
> The only question is if the rest of the world is going to wake up and counter this.
Probably not.
Simply put, there's no first-world country in the world where top talent can earn even close to a US salary, which largely due to the fact that US is relatively permissive towards large income disparities.
Other countries "fixing" that would be politically infeasible.
> Simply put, there's no first-world country in the world where top talent can earn even close to a US salary, which largely due to the fact that US is relatively permissive towards large income disparities.
Why do you think it's A -> B and not B -> A? What factors of the US being permissive towards large income disparities do other nations lack that prevent them from paying equivalent salaries.
One specific thing I've heard which makes sense to me toward explaining the salary gap is the US tends to have at-will employment while other nations make it more difficult to fire people. I wouldn't personally classify as promoting income disparity though.
It has an undeniable ability to draw the best talent away from Europe and Asia, who want to earn more. I have worked with armies of these people -- they left their home countries for the opportunity for the big check. And those checks are way past the disparity you describe from at will employment, this is 10x more not 1.4x more.
This is why open borders and immigration is a right wing capitalist value and not a left wing one. Many left wing activists have fallen into this trap of towing the line of giant US corporations that strip the world of their human resource.
> largely due to the fact that US is relatively permissive towards large income disparities.
Actually, its due to a shortage of highly skilled labor/specialists relative to demand. Europe benefits from more advanced math and programming talent as a result of Eastern European and Russian/Ukraine labor availability. US immigration policy puts America at a disadvantage in this regard - something they are weakly trying to address here.
Sorry if that goes against your income inequality agenda.
> Not sure why the barrier to using Eastern European talent would be appreciable different.
I mean, Poland, Czech Republic, Hungary, Romania and Bulgaria are all EU members, which means they have freedom of movement - they don't even require any visa or work permit. Just show up in Western Europe and work. The same is currently temporarily true for Ukrainians.
Compare that to the extreme effort for them to get a green card in the US.
Don’t you think if this were true there would be more breakout technical (and by extension startup) successes in Europe which doesn’t seem to be the case.
Yet at the same time Liberal thought is the reason we have made such an incredible break thought in AI technology with the LLM. Quite literally the hippy movement lead to the open source movement lead to the internet lead to the LLM. So we have this seeming conundrum between little l and big L liberalism...
... im struggling with the realization that Liberalism the big idea is probably an afforded luxury ideal of the rich, as well as a runaway self-re-enforcing flywheel of wealth/advancement. I benefit from it, but fewer and fewer are going forward, while the US is simultaneously winning harder and harder. This is a nightmarish equation!
At the same time, moving to the US is unattractive for a significant portion of the talent (not the least due to the US’ socio-cultural-political climate), and working remotely for a US company comes with its own difficulties.
Haha this is such a globally out of touch statement. Yes a few European nations enjoy a much better quality of life while looking down their noses at the rest of us, blocking immigration, and generally being xenophobic.
And it’s not like the climate in Europe isn’t verging on ethno fascism in several major state…
I suggest that a substantial portion of that "socio-cultural-political climate" is due to foreign influence operations trying to amplify division for this purpose.
I'm from Finland myself, I've been living in the US for some years, and I have no concrete plans to return. I've known many Finns who have tried the same, but most of them have returned to Finland, or at least to Europe. Most of them are highly educated professionals. It turns out the difference in standard of living is not that large. For most people, living closer to family and friends is more important. If they find a decent opportunity closer to home, they usually take it.
You can see the same effect with Finnish nurses. They could improve their standard of living significantly by moving to Norway. Probably more than software engineers moving to the US. It would be easy, because Norway is close, the culture is pretty similar, and Finnish citizens have a subjective right to live and work there. But few people actually do that.
There are definitely small ideological, well-off pockets of Europeans, who enjoy living in high safety net countries and don't really want to trade one developed country for another.
They don't necessarily have the skills to draw high US tech salaries either.
But for the vast majority, like you said, the US is extremely attractive.
> "America's global brain drain might have been the single most effective strategy"
The website has a section titled "Bring your AI Skills to the U.S.", and then a button with this label "Learn about pathways to work in the U.S.".
This is very encouraging. And it reminded me of Eric Schmidt's recommendation on this topic [1], which was published on the final report of the National Security Commission on Artificial Intelligence:
Chapter 10. Page 178:
"Nations that can successfully attract and retain highly skilled individuals gain strategic and economic advantages over competitors." [...] "Unfortunately, international students in the United States are increasingly choosing to study in other countries or return home. One reason is the growing backlog of green card petitions. Indian immigrants face a particularly long wait. Many will spend decades on constrictive work visas waiting to receive their green cards, hindering both the technology sector’s ability to recruit talent and Indian immigrants’ quality of life."
The report recommends to focus on building better Immigration Policies. Maybe this .gov initiative listened to the advice from the report.
Yea I think it's more that Schmidt believes that AI is the force-multiplier of any existing system of power in the world today, and whatever it takes the US needs to continue to fuel it.
But he also seems to understand the labor/intelligence relationship to get there better than anyone else. I really missed him when he left Google, Sundar is a pail shadow.
I find that very interesting given Eric Schmidt took citizenship in Cyprus, and to my knowledge, not because he has any prior tie to that country. If he really believed in the U.S. then he shouldn’t be getting late-in-life (post-billionaire) citizenship elsewhere.
No one else is really capable of making this technology at the level of the US, besides maybe China. Europe has the brainpower for sure, but they don't seem to be able to leverage it into actual innovation.
> The US is now on the verge of the second great breakthrough of mass automation of that data, intelligence and action. I am honestly shocked that this third stage of American Dominance has not be countered by any other nation save China, and even they're only defensively.
As someone who has similar thoughts, I'm not so sure. The EU has been trying for a while. Russia also puts quite a bit of effort into having an Internet that can operate separately from the US.
edit: That still doesn't explain to me why China and Russia don't have stronger pro-Linux (or something) initiatives. Even the middle powers could at least try to have some control over their technological infrastructure.
I think the foreign influence operations that aim to increase division in American society and promote equality of outcome are an attempt to counter it, not by outcompeting America but by hobbling it, seem to be somewhat effective.
Yea that seems to be the idea -- try to constrain America as much as possible if you can't beat it on its own terms. However US military as grown such that it likely can fight a 3 front war and most other nations would rather the sweet sweet trade deals they get by cooperating over some obscure ideological vision of a world without america.
There's good reason I run GrapheneOS and Qubes OS exclusively on my devices, and I route everything coming out of my systems over Proton and Tor (ofc not nested xD)...
The Presidential Innovation Fellow is the highest level role they're offering as part of this. For this, the requirements are "mid-senior/executive career professional experience in a subject matter expertise" with "experience collaborating with and leading cross-functional teams".
The requirements feel analogous to a staff/principal level role in the software industry, so just for comparison here's the compensation for these roles across the giants:
Microsoft (level 66): $390,707
Amazon (L7): $547,490
Facebook (E6): $610,713
Apple (ICT5): $497,780
Google (L6): $506,141
US Gov: $155,700 to start, with a max of $183,500 after 18 years (step increases are limited on a time basis)
The US government is trying to leverage AI by paying 30.4% the wage of the leaders in private industry. I just don't see how they're going to be effective in doing this.
I assume they're targeting people who aren't competitive candidates for big high paying companies. There's plenty of super mediocre to bad developers out there (myself included) that would never stand a chance applying there. I've been developing software for a decade and the most I've ever made is $110k, and that was in high cost of living areas like Denver and Seattle.
How come? I would value a Stanford degree and working at Google on a resume. I wouldn’t know what to think about The White House or Air Force (like many posting on their site) on a resume.
It's specifically the Presidential or White House [0] fellowships that are prestigious, although if you have any job in the White House that's just as well.
What value an employer would derive from seeing that on someone's resume is probably the same as with Google or Stanford: it shows they got in, with the known/assumed quality of the institution vouching for their capability.
The pay cap is set in law, but you need people sufficiently invested in the mission to forgo a higher possible salary, which isn't possible for everyone (especially folks who might have to support their parents).
US gov hiring is absolutely broken. I applied for a role I'm absolutely qualified for (even referred to apply for it by someone on the team based on my resume and experience) and was rejected out of hand (twice, after an appeal too) without consideration. I'm actually willing to take a pay cut to serve my country, and I think actually you'll find senior engineers and business leaders willing to do this too, but if red tape or incompetent HR contractors get in the way, you'll lose that talent too.
- your 18 year figure is inaccurate, I know people who have made it to the top of the pay scale in half that time
- GS-15 step 10 is not the cap for a role like this one. this is scientific/technical expertise, which is 20% higher ($220,200). Not to mention awards.
The Executive Schedule is available for 18F as well as other agencies, but my understanding is that it is not true for this. Screenshot from the job page itself showing that this is a GS job capped at the GS-15 Step 10: https://image.non.io/6dbd428b-a356-4ec9-85a8-f1e84b4f3880.we...
Yeah but your country needs you, and that’s actually pretty good pay. I think tech would be a little bit broader if more people considered serving for a year or two.
Who says you can get more done for the country in this role than you could working in a research lab at a tech company? Certainly US advances at companies do A LOT for the country and it's HARD to get anything done in the government. So, for the top AI talent this appears to target, this is the opportunity to get paid way less to accomplish less..... Why would anyone do that?
Countries are like a streaming service. Shop around to see which you like most and subscribe to it in the form of paying taxes. If a country has brain drain then that’s their fault for being incapable of providing a good life to their citizens and inherently should not exist as it means they’re doing their citizens a disservice by providing a worse quality of life.
The cost of living varies widely in different areas. In many parts of the United States, 100k is a very low salary, due to extremely high living costs.
For example, the government classifies a family of four with an income of $149,100 as "low income" in San Francisco, and thus eligible for government welfare benefits.[1]
The median house price in San Francisco is currently $1.3 million.[2]
An average 776 square foot apartment (which is rather small for a family of four) rents for $3,336 a month.[3]
> After 100k, what difference does it make for one's happiness?
Fairly significant depending on the location. Keep in mind that for Washington DC, the average cost of living required income is 91,388 for a family of four[0], or 76,894 after taxes.
This role pays $105k after taxes. One way to think about this is this means you have $2.3k per month in discretionary funds, versus 16.5k per month of discretionary funds for a big 5 job.
If you're considering retirement, you'll achieve your savings goals 7 times faster with big tech. That's a big difference in happiness.
If you're considering education for your children, you can afford private school or an ivy league education with big tech. With the government role you can't.
Certainly there are a select few who will look at this and be motivated to serve their country, but more and more what we're seeing is that patriotism is at a record low[1]. If we're looking to attract the best of the best, we're culling the selection pool greatly by underpaying significantly. Relying on patriotism and asceticism is a hard thing to bet on.
100k is very little.. I’d want my government workers to have enough disposable income to frequently travel abroad and experience different ways of life.
Yeah, you don't have to spend it. Save it and retire early. Donate it all to a charity you believe in. Send your kids to the best school. Buy a nice home. Travel the world. The high salary only allows more options, it doesn't close any doors.
(Just be careful about "getting used to it" and ending up having to change your lifestyle if market conditions change.)
Your comment is a cynical one. The very people qualified to define something like AI safety for the US population is...the US Government. Which is what they're doing. If you don't like some of what they're doing, there's many pathways to making sure your voice is heard, including voting for people who would explicitly support your policy preferences.
>nobody actually knows what that means or is qualified to define it yet
Now is the time to start talking about it and taking it seriously though. If we keep avoiding trying to define it then we'll never be qualified to do so.
I realize you're making a direct reference but scare-quotes seem more applicable.
https://en.wikipedia.org/wiki/Five_Eyes
"documents of the FVEY have shown that they are intentionally spying on one another's citizens and sharing the collected information with each other"
What else would you expect? Both issues are about power, and I say that in the most matter-of-fact and non-judgemental fashion I can. The US government will effectively defend the privacy of American people as long as it poses a vulneravility, and that is one facet of AI safety. But a "right to privacy" as a domestic matter is different beast and there's clear malalignment.
On their "immigrate" page ( https://ai.gov/immigrate/ ) , they say "We must attract, train, and retain the most talented workforce in the world," and mention H-1Bs as a part of that solution. But federal law caps H-1Bs at 60,000 a year, of which over a quarter goes right to Amazon and AWS (EDIT: Turns out I'm wrong about this!). Does the government as an employer get to circumvent the cap? Or are they suggesting that a larger share should be used for AI stuff? Or are they advocating raising the cap? Or is it just a meaningless website?
It is a meaningless website. Everybody and their dog knows that in the present political climate it’s practically impossible to change the immigration law. It’s only a remote theoretical possibility that the US could fix their immigration system.
No idea how reliable that site is, but it says Amazon.com sponsored 16,299, Amazon Web Services sponsored 5,056, and Amazon Development Center sponsored 2,119. So 23,474 in total.
Edit: I see a note that this is APPLICATIONS, not grants. Nevermind, that's probably not accurate.
Those LCA applications aren’t 1-1 with H1B quota numbers. Eg after 3 years when renewing a H1B the employer needs to file an LCA but that won’t count against the cap. Similar for H1B transfers and renewals past 6 years.
Overall the claim that Amazon takes 25 percent of H1Bs is extremely incorrect.
I believe there is no publicly available data that allows one to calculate the exact percent of new H-1B numbers used by a specific company.
But if I go into USCIS's H-1 data hub, I can find some information. For FY-2022 Amazon had 12.5k petitions approved. But drilling into the data we see that less than 4k of those were for _new_ petitions - the others were extensions of existing petitions. 4k out of the H-1B quota is less than 5% (there are 85k new H-1B visas per year). But even that is an overestimate because a new employee transferring to Amazon will require a new petition but won't count against the quota.
Overall, the best guess would be that Amazon use low single digits percent.
Occasionally I enjoy viewing the HTML of websites to try to deduce what web framework they are using. I couldn't figure it out for this one, but here are some things that I find interesting:
- On line 90 of the homepage, there is a single script called `document.createElement('main');`
- The website uses the `<picture />` tag to support multiple sources for images
I cracked open an internal tool we had one of the big consulting firms build for us to see why it sucked so much. In one of the header elements they had hardcoded the URL for a QA site they built for a large car dealer. Fun stuff.
Interesting that on the page of AI use cases[1], as well as the "full list" CSV[2], Department of Defense is simply left out. I guess they're happy to advertise all the Good Uses of AI, but don't want to publicize the Bad Uses that are definitely being pursued.
Off topic, my initial reaction was it would have been more of a flex if they managed to get the domain gov.ai, since US gov already owns the .gov TLD, but it turns out that this domain is already used by government of Anguilla which already owns the ai TLD.
Looking into this further, I also learnt that in fact selling ai TLD has become a non-trivial part of Anguilla's GDP since chatgpt was released.
Even if it wasn't already used, putting putting official government stuff somewhere that's under the control of a different government is maybe not the best of plans.
I'm even mildly annoyed that the website for the city I live in is under .org rather than under .${STATE_CODE}.us where it should be.
Oh, look at the tortured path that .io dollars take, since it's technically a TLD for an occupied country, being operated by someone (Ethos Capital) with no connection to or permission from the original or current inhabitants.
Available yes, but many are "premium/aftermarket" pricing now. There is such a high proportion of these names/prices popping up when you do domain search compared to the dominant "Unavailable" of .io or .com.
Source: I've registered .ai names since 2019. I use a 5 and 8 letter fairly common english word for two .ai domains; both were standard pricing originally, but comparable available domains are now >$10k. However, I could probably at best resell my domains for < $1k each, as resell value for .ai is hard once you use them. I guess investor dollars and being seen to do something might due the demand for these premium domain options.
I think the grandparent comment was talking about making AI Force a separate military branch, rather than a joint office under DoD.
Not arguing one way or another myself, I think it makes sense for it, at least for now, to be a joint office under DoD (like it currently is, according to your link). Just wanted to make it clear that CDAO is probably not what the grandparent poster had in mind when they made that comment.
I was a Presidential Innovation Fellow for almost three years at FDA / Veterans Affairs, happy to answer any questions. The program is looking for industry experiences to better inform and shape government policies.
Serving at USDS, posting personally. The biggest thing with the gov is the requirements on usajobs are always explained specifically so you know exactly what is needed.
Beyond that, you'll want to show in your resume how you've delivered successfully and ideally in challenging circumstances, because working in the gov can be challenging. It will really vary depending on the job and agency.
To echo wslack, experiences with moving initiatives through grit and collegiality is key. The PIF program is a bit easier to apply for than USAjobs and really focuses on being more of an entrepreneur-in-residence, as much as that’s possible in government. I was able to work horizontally in a way that would have been challenging otherwise.
Lots of folks commenting about salary but that's not even the biggest drawback here. Org instability is.
The presidential innovation fellows are 1-year stints, and after the next election, there might not be a next role. Maybe a campaign? Maybe a consulting firm? Unclear.
Even if the parties don't flip - senior leaders churn through fast.
I love this attempt, but I can't imagine trying to be a mid-senior tech leader in an organization where, in 18 months, the next guy might be trying to actively tear down your work.
I mean, that's an issue with administration turn-over in general.
With the PIF it looks like they get spread around to various agencies, so I don't think a fellow from one cohort is likely to be assigned to the same agency and project as a previous one.
There are a lot of folks interested in or concerned with what the government is doing here. I would just mention that going to government on a lower salary isn't a binary thing and you can do impactful work for 1-3 years. Places like TechCongress, Presidential Innovation Fellows, and myriad other fellowships have popped up to support this.
My friend,
it is quite literally the goverment's job to decide what is right or wrong through drafting and enacting laws.
I am not stating that having to inform the government when you're training models is right. It's an asinine way of thinking by people who have no idea what they're talking about and are severely out of touch. The laws they've enacted were based on fear, often propagated by OpenAI themselves. It is still their job to do that though.
I would really love to work for the USDS; ever since their big publicity outreach during the obama admin, too bad the salaries are abysmal comparably to most tech companies. 184k top band in san francisco; 150k in DC. I could join a mid tier firm in SF for 250k base; not even bonus for a senior level. I wish they could create a special schedule.
Silly question - but what do people in AI related roles do? Are they looking for software engineers to use/integrate AI products such as OpenAI APIs or are they looking for researchers working on AI papers trying to break new ground?
I predict the rise of anti-technology cults as the moral panic over AI and “Social” takes hold. The millions of displaced workers and disillusioned social-users will need refuge.
I predict that AI is going to increase demand for workers in knowledge work, though it will also change the nature of some of that work.
When an efficiency increasing technology comes along, things tend to grow and not shrink. Cars massively increased the number of people employed in transportation related and adjacent fields, to give just one example.
Social is another beast entirely. There's already a huge backlash there, and justifiably so.
I believe we're witnessing a political realignment, and anti-technology sentiments play a significant role in the evolving left/right ideologies. There's something I term the "new right," which is distinctly different from the "alt-right." The new right leans more libertarian and has a younger demographic.
This group values individual liberty in financial matters (e.g., cryptocurrency), individual liberty in computing (e.g., AI), and individual liberty in manufacturing and firearms (e.g., 3D printers and semi-automatic guns). Interestingly, they also advocate for transparency regarding extraterrestrial life. It's an unusual stance, but it's part of their ideology. Of course many prominant democrats have led the charge on this, so it's probably not as cut and dry like the others. During the COVID-19 pandemic, they were skeptical of mainstream narratives, but their views were diverse; they weren't strictly anti-vaccine or pro-vaccine, anti-mask, or pro-mask. They harbored various opinions, but universally believed that the experts' confidence was misplaced.
In contrast, the opposing side seems to advocate for stringent regulations on AI, cryptocurrency, and firearms. They seldom discuss extraterrestrials, primarily because they view themselves as experts, and in their eyes, the existence of aliens is implausible.
The "new moderate" recognizes the potential of AI to cause widespread unemployment but also understands its immense value. They believe that a balance must be struck. The new moderate acknowledges the benefits of cryptocurrency, especially in regions lacking a stable banking system, and the potential of smart contracts. However, they also recognize its misuse by terrorists and its frequent use in schemes that prey on uninformed investors. They believe that compromise is essential, and some form of regulation is inevitable.
1) No cloud, everything local or site-to-site. If cloud, then encrypted entirely.
2) Not only FOSS operating system, but FOSS firmware and microcode, and the hardware completely documented and published. Maybe a 100% sales tax on retail price of, and 25 year software service assurances on, products containing ANY kind of Universal Machine which fail to meet this?
(Elect me Benevolent Dictator For Life and I will make this happen!)
3) Cryptography and cybersecurity basics need to be taught in public schools, just like resumé writing.
If you disagree with something the President or one of his subordinate agencies does, write your representatives in Congress and ask them to pass legislation. The President can't do anything that Congress tells him not to do.
Good news - you're not, because your tax dollars doesn't pay for much of anything. The US Government has monetary sovereignty and "pays for things" that way.
Imagine if you had clicked the link and learned that this was a site for finding AI-based Government jobs, and not some sort of regulatory statement where the government was planning to "ban AI models".
But that would require basic literacy, something that conservative reactionaries simply don't have.
Likely a right-winger who thinks immigration - even immigration of people who are talented in developing the most important technology of our time - is really bad. So they "couldn't resist" offering more immigration opportunities to people.
Look at his post history. He recently complained that LLMs are too "woke".
So yeah, 100% a racist right winger who didn't have anything of value to say, and decided to just throw shade at immigrants when he saw the opportunity.
Good news is that these guys will open their mouths at work and get themselves canned, so we usually only have to worry about them online.
A perfectly non-racist statement in a world where context isn't important. A very racist statement in a world where everyone knows that you're saying that foreigners are dangerous.
The stuff on the page is already known. They just add it to cross off a checkbox of what they wanted to signal. And its just a generic description of the visas and nothing localized to th visa type and ai.
And for the other commenters calling me "right wing', my views are pretty much '04 Obama. It's just telling how far off the rails we have gone.
Also, nothing on that page was factually incorrect, so we come back around to the original point: Immigrants were brought up, you hate foreigners, and you got mad that they were included at all.
The US really needs a special schedule for software developers if they want technical expertise in house (and they really should have technical expertise in house)