Slashdot Mirror
Tech Giants Are Paying Huge Salaries For Scarce AI Talent (santafenewmexican.com)
jmcbain writes: Machine learning and artificial intelligence skills are in hot demand right now, and it's driving up the already-high salaries in Silicon Valley. "Tech's biggest companies are placing huge bets on artificial intelligence (Warning: may be paywalled; alternative source)," reports the New York Times, and "typical AI specialists, including both Ph.D.s fresh out of school and people with less education and just a few years of experience, can be paid from $300,000 to $500,000 a year or more in salary and company stock." The New York Times notes there are several catalysts for rocketing salaries that all come down to supply and demand. There is competition among the giant companies (e.g. Google, Facebook, and Uber) as well as the automative companies wanting help with self-driving cars. However, the biggest issue is the supply: "Most of all, there is a shortage of talent, and the big companies are trying to land as much of it as they can. Solving tough A.I. problems is not like building the flavor-of-the-month smartphone app. In the entire world, fewer than 10,000 people have the skills necessary to tackle serious artificial intelligence research, according to Element AI, an independent lab in Montreal."
I am sure "Element AI" wants to pretend there is such a thing as AI, but there isn't. Playing "Go" is not "AI" and neither is autonomous driving. If you are going to start calling computer algorithms and programs "AI" then everything that runs on computer "AI".
A sufficiently talented AI 'researcher' should be able to code a self-aware AI researching AI.
We don't have strong AI yet. There aren't 10,000 people that know to get to strong AI, there are currently 0. For our current AI, anyone can develop it, and you just need marketing to announce it as the next big thing.
Since AI is the be all and end all, they should have their existing AI geniuses write some awesome AI logic that does the same thinking and work of other AI geniuses. Problem solved.
It's a good thing we have a president who is focused on science and education. Now we'll never lose to foreign countries!
In the entire world, fewer than 10,000 people have the skills necessary to tackle serious artificial intelligence research, according to Element AI, an independent lab in Montreal
In the entire world, fewer than 1000 people have the skills necessary to do unstructured tetrahedral finite element mesh generation. It is possible there are fewer than 1000 people who have the skills necessary to understand what exactly we mesh makers do. And, Surprise! there is demand for fewer than 1000 people to write unstructured tetrahedral finite element mesh generation. And far fewer than 1000 people are needed to manage them.
I am glad the periodical bubbles that infect Wall Street and venture capitalists benefits PhDs once in a while. Most of the time it benefits hedge fund monkeys or stock market cheats or lottery winners with delusions of grandeur or plain sociopaths. Happy for my grad school classmates. Enjoy the windfall while lasts, Ramachandran\s, Yang\s, Hsu\s, Gupta\s, Parpia\s and Wickramasinghe\s.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
Greetings Professor Falken!
Hello [UserName], my name is doctor Sbaitso.
I am here to help you.
When companies start paying like that, that's indicative of shortage.
Remember this folks the next time some company claims they can't get enough qualified people.
I just finished my PhD in neural networks and AI, and I can say that this article comes up a bit short. I actually got offered $625K (granted it's San Jose, so it's more like $150K anywhere else in the country)... I turned it down because San Jose sucks, so I took a job at a company on the east coast instead for $195K.
AI-related jobs are certainly at a premium right now as companies scramble to get rid of humans and replace them with robots.
Modern AI software isn't that complicated and not nearly as expensive to get people in. Look at job offers: $150k for AI research scientists in NYC. $65k in more rural areas. That's not well paid by definition at all. Sure, a pure AI scientist gets paid $500k just like a top neuroscience scientist gets paid $500k or a top biology researcher, but the majority of companies do not want to do the theoretical development of AI, any regular programmer can wrap their heads around the existing literature and build something.
Here in my area, there are a number of employers looking for AI engineers/scientists. They pay about what I make as a non-AI IT sysadmin, which is given my experience on the higher scale but by no means exceptional.
What Google and co wants is a glut of people 4-6 years from now that are "trained" in AI from college. You put out a report like this, you get massive amounts of people applying for the schools that offer programs and 5 years from now you have an over-abundance of people driving down overall wages. You also get to hire a bunch of people on H1B because the "US doesn't have the skillz" and you end up with a bunch of programmers on H1B under the guise of AI development.
Custom electronics and digital signage for your business: www.evcircuits.com
This is spin. They are trying to get more entrants into the field. I am in the field, and I don't know anyone making more than ~120k/year, which is mean for CS workers in Silicon Valley.
It is just one more AI-generated Slashdot headline!
Since I have scarce AI talent, does that mean I can get a huge salary? I wonder what they're paying people with ample AI talent.
I just got mine in gender studies and I have been writing papers on how AI is the projection of masculine hegemony, domination, and misogyny into the computing sphere. And how it will add to global warming and increase the relevance of the penis social construct in masculine psychology and subjugating other genders including trans and cis genders.
It's a great job. I write gibberish, teach classes to sanctimonious well-to-do white kids, and get paid almost 6 figures - for working about 4 hours a week.
Well, at least we'll have another bubble to form and expand once the bitcoin bursts. It seems like this is the way the stock market works. The few at the top profit heavily from the bubble, the bubble bursts, and the 99% are left to deal with the aftermath.
From my investigation of the matter it looks to be some sort of multi-variate analysis in drag. Uninteresting. Basically you get guys sitting around twiddling knobs. Finding the right parameters which works for a little bit and then you start knob twiddling again to find the next ones.
Except for 1 key difference. With Deep-Neural-Nets, the knobs twiddle themselves alone.
DNN get inspiration of how some neural network work in the nature (e.g.: a column in the primary visual cortex of the brain) to design thing that you can throw at problems, and which will autonomously train themselves.
Some years back I wrote a day trading program for a friend. It dynamically changed its behavior depending on the market signals and the rules he gave it (stops, buys, shift to a different stock etc.) which he found useful. Now that was fun.
These older program require you to have precise criteria in advance.
That works perfectly well with clearly codified problem - the friend has a clear set of rules that need implementation.
That completely fails for more vague problem ("detect a face") - it would be possible in theory to design a set of rules that can detect a face - a Haar Cascade. But designing such set of rules is extremely complex and cumbersome. And each time you need something new ("detect if there's a bird"), you would need to repeat all the hard work to invent yet another set of rules.
At that point, better take an advice from how mother nature solved the problem (by using stacks of neural network in a columns) and simply throw a DNN (e.g: a Convolution Neural Net - a ConvNet) at the problem, and watch it self organize and come up with a solution to your problem.
It's the modern-day equivalent of training pigeons to peck a city images to steer a missile.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
You won this !
What these tech giants may find, it is very difficult to find someone with clearance and necessary skill set. Many current government contractors will hire someone with the clearance and provide the training for the skills.
Not really. Machine learning and data mining is a subset of what AI would be. Real AI would require a much bigger skill set not usually encompassed in one person.
It's just collecting the data, which usually involves a lot of manual as well as automated processes, and then crunching and analyzing them using automated means.
BTW we saw how well 538 did predicting last November's election. Yeah, yeah, they weren't "wrong" because they did say Trump had a 23 percent (or whatever) chance of winning on the eve of the election. Michael Moore knew better.
""Tech's biggest companies are placing huge bets on artificial intelligence (Warning: may be paywalled;"
And also no paywall, if you delete your cookies after the allowed number of articles.
These 'paywalls' are a joke.
Oh there are always going to be bubbles. Just look at all those people trying to own a computer. When that pops, disaster.
Congrats! You must not live in a shitty country like Russia. Post away!
Nice rhetoric—factual statement masquerading as metaphor, for any reader dumb enough to go along for the ride.
The Evolution of the Flour Mill from Prehistoric Ages to Modern Times — 1905
That's about the present state of machine learning, the hand-crafting of "features" playing the role of the recently discarded flat blocks.
Wheat is an incredible dietary resource, with the starch being light enough to transport over long distances, if only one can find a way to remove it (contrast potatoes, only ever transported downhill, if at all, until the invention of steam power). Once upon a time, all food was local, as, too, was starvation (fear the blight).
A better method to mill the world's vast stores of accumulated data is a big deal, even if we remain in the relatively crude era of water-powered stone grinding wheels.
Data is a bit like wheat, it doesn't give up its curvature easily. Too much applied force creates heat and destroys the end product. The applied force must have exactly the right ratio of compressive to shear stress, which only an expert miller can judge. Deep learning is nothing more than a slightly better mill than the one we had before, and it ranks right up there beside becoming slightly better at milling wheat.
The economic value of the curvature we can now hope to unlock is quite large. And probably there's a lot of curvature yet to find that remains inaccessible to current methodology.
Data is oil. Data is also wheat.
By way of contrast, unstructured tetrahedral finite element mesh generation shaves 5% of the metal mass off a milling apparatus that already worked just fine, being just one of ten thousand noisy specializations in the great roil of small improvements where a penny shaved is a penny earned.
Nevertheless, apparently a great career option for the metaphorically challenged.
to what is being enabled ;)
Sure, right now you need a PhD to get one of these jobs, but have no fear, in another year someone will create an opensource JavaScript AI library and then, provided you have 10 years experience using this, you'll be fending off bids from every firm in the valley.
You must not be at a US school. Most of us now are adjuncts, making less than public school teachers.
I know a guy who cut and pastes Javascript snippets to make interactive query windows for websites. His main job is as a creative director at a mid-tier advertising agency. He calls his work 'AI research'. I am not kidding - he makes over $100k per year.
The biggest problem I have found with smart people, is they don't think stupid people should be paid lots of money for work they think is simple. The more successful ones have figured out that it is much better to cash in on such situations rather than lament what the world has come too.
Author writes "AI Talent", then refers to Machine Learning.
Machine Learning refers to a few statistical regression techniques that is not what artificial intelligence is about.
Artificial Intelligence is more of a research field, and the concept of human intelligence remains an unsolved problem ---
what it sounds like they are really hiring are hard-core Computer Scientists with some experience attacking real-world solution-finding problems like extracting useful intelligence from data, classifying or identifying things, making decisions, or acting in the real-world; that goes outside the bound of AI, because the common goals are to make computers solve real-world problems for business tasks and Not make generally intelligent machines with agency and capabilities similar to a living intelligent being.......
$300,000 to $500,000 is not huge salary, it's considered low income. Remember ?
I would like to pose a different perspective. Yes you are correct in this being an algorithm approach which is being called A.I. I think though this is a matter of speaking what doesn't exist into existance by placing focus, and as a result, resources, and as a result problem solving, and most importantly determination. Now there are people like you who glory in pointing out the discrepancy, kudos.
car (or missile in this case. Same thing)
https://www.youtube.com/watch?v=bZe5J8SVCYQ
And in simpler mathematics here.
Epic.
No one capable of teaching these experts, because everyone is still learning, but lets pay them 500k a year instead of trying to figure out the best algorithm ourselves - we're the experts in our field after all - why waste time on learning something easily marketed currently?
I see you, scumbags.
AI in today's IT world is a misnomer. There is no true AI only cleverly designed exceptions lists.
They should hire me for a huge salary...
I did an AI course as part of my BSc Computer Science back in the 90's.
For a project I made an "Expert System" (or was it called a Smart System back then), wherein I wrote a front end in VB6 (might have been 4), that was attached to a Access DB that contained all the beers on tap (I think there were 30ish) at one of our favorite drinking establishments, along with all of their characteristics, that prompted the user with a bevy of questions to determine and suggest what the optimal beer that person should order.
I did get like a 97 or 98 percent on it... :)
I bet you spent a good bit of time on the "testing phase" to make sure it really worked.
I only look human.
My mother is a halfling and my dad is an ogre, so that makes me an Ogreling
Why is scaling a big thing. If it works with N parameters, it will work with N+1.
One of the reasons is complexity of what the parameters describe and how they can be understood by the researcher.
The way it was done in aeronautics, could be compared to piloting a helicopter :
- You have a position that you want to keep (easily described with 6 number giving coordinate and direction pointed to).
- You have a bunch of controls (Cyclic, Collective, Anti-Torque, Throttle) with each subtly influencing each-other because of gyro effect.
You optimise the main parameters (controls), and maybe you have an indirectly layer of a few implicit parameters as well (Yaw, Pitch, Roll, Raise rate) with controls not only directly but also indirectly mapping to them.
Nearly everything of the above makes sense to the researcher.
Deep neural nets could be considered the same, but turned up not only to 11, but to 10^11.
You have a picture, you have a dozen of layer, you get a bunch of parameters. You can use said parameters as a signature of the object looked at.
You can give some general signification of the first few layers. (contrast/edge extraction, lines/directions extractions, texture extractions, etc...) but no single value make any sense by it self (compare to the Yaw, Pitch, Roll, Raise rate) )
Beyond this few layers things start to go really banana.
Its hard to put an exact meaning to any single cell of this huge network. And you don't even need to consider them. The cells will find each their own internal meaning alone during the training.
You are basically doing the in-silico equivalent of training pigeons to peck at picture of cities.
You have a general idea of how an animal's visual system works. You can even go poke inside its brain with electrode to measure local variation.
But you can't really comprehend what every single synapse between brain's neuron are in charge of. At this scale it becomes meaningless.
Instead you count on some general forms of test to make sure that process works as it should.
(random example : you can tweak noise until is makes a pre-trained system light up. The noise will become a weird surealist distorted hallucination of what you trained your neural net for - search google image for dog nebula).
The sheer scale of the stuff make it meaningless to comprehend in details. You consider it at a higher level.
You don't think about logical gate of silicon transistors when you think about information flow across the internet.
Even if the internet is an interconnection of network, which in turn are interconnected computer, which in trun contains lots of chips, which in turn a composed of myriads of logic gates.
But nobody thinks about the logic gates when solving internet problems.
And the same way nobody thinks about single parameters when considering deep-neural-nets.
The second thing is that at this scale, you take a lot of shortcut to propagate the "signal" (i.e.: how each parameter influence the next, how each cell of the network is connected to the next one). And you train the thing until it more or less produce acceptable result, not until it stays stable. Otherwise you would be asymptotically approaching a wall where there is simply not enough existing computing power to throw at it.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
Americans are addicted to defined benefits. Defined benefits, however, were always a scam, and are still a scam, as they are mathematically impossible to sustain for long periods of time, depending upon bad assumptions and wishful thinking.
Corporate America has finally come to realize that the only way to get rid of defined benefits is to get rid of the people. Get rid of the people, you get rid of the unions. Get rid of the unions, and you get rid of defined benefits - at least most of them.
Corporate America doesn't care if 90% of the country lives in poverty as long as the remaining 10% generates them profit. Government certainly loves poverty because poverty is what cements it with power.
You are all a bunch of tools if you don't see this.
AI & ML have EXPTIME and EXPSPACE problems: not found NP = P for these things of AI & ML.
They want fastest algorithms for short time & space that are very difficult to find.
While this may just be a bubble, I'm hoping it's one I can get in on. My Master's degree in AI isn't too far off.