'We Can't Compete': Universities Are Losing Their Best AI Scientists (theguardian.com) 268
The Guardian shares the story of a PhD student at Imperial College London who abruptly stopped coming to the facility, even as he had one-year of studies left. From the story: Eventually, the professor called him. He had left for a six-figure salary at Apple. "He was offered such a huge amount of money that he simply stopped everything and left," said Maja Pantic, professor of affective and behavioural computing at Imperial. "It's five times the salary I can offer. It's unbelievable. We cannot compete." It is not an isolated case, the report says. Adding: Across the country, talented computer scientists are being lured from academia by private sector offers that are hard to turn down. According to a Guardian survey of Britain's top ranking research universities, tech firms are hiring AI experts at a prodigious rate, fuelling a brain drain that has already hit research and teaching. One university executive warned of a "missing generation" of academics who would normally teach students and be the creative force behind research projects. The impact of the brain drain may reach far beyond academia. Pantic said the majority of top AI researchers moved to a handful of companies, meaning their skills and experience were not shared through society. "That's a problem because only a diffusion of innovation, rather than its concentration into just a few companies, can mitigate the dramatic disruptions and negative effects that AI may bring about."
Surprised? (Score:5, Insightful)
Why should they be surprised? PHDs are treated like crappy free labor by universities.
Perhaps when they stopping paying administration officials obscene salaries and pay professors and grads what they are actually worth the quality at universities will improve.
Re: (Score:2)
Re:Surprised? (Score:5, Insightful)
Perhaps when they stop paying football and basketball coaches obscene salaries and pay professors and grads what they are actually worth the quality at universities will improve.
Fixed it for you.
Re:Surprised? (Score:5, Interesting)
Perhaps when they stop paying football and basketball coaches obscene salaries and pay professors and grads what they are actually worth the quality at universities will improve.
How do you gauge what is obscene and what people are actually worth? For instance, back in 2010, it was estimated that Tom Izzo was paid $3 million, but the basketball program generated roughly $11 million for the university [espn.com]. So do you think the professors and grads should be paid 27% of the revenue they generate for universities? How do you calculate that? What if they don't generate any revenue? Do you pay them nothing?
Re: (Score:3)
$11 million "for the university" -- like somebody walked over a check to the CompSci office and said "Here, give those AI researchers a raise!"
It made sense on paper for big athletic departments and sports programs to become "self-funding" -- they could spend whatever fans and boosters wanted, so long as they raised the money to pay for it without spending University money.
The problem was, if they were actually profitable they began to claim their own profits either outright or by inflating their budgets ra
Re: (Score:2)
Perhaps when they stop paying football and basketball coaches obscene salaries and pay professors and grads what they are actually worth the quality at universities will improve.
Fixed it for you.
You're making the assumption that hiring a different football coach and paying them $1-2 million less will not have any effect on the revenue generated by the football team. I'd be interested in seeing your reasoning or evidence for making that assumption.
Re: (Score:2)
There is no "reasoning" here, it is all emotional rhetoric of the uninformed. All they see is "$X Million salary" and get all envious without understanding that salary allows revenue that funds all sorts of things and keeps University Boosters donating to various programs. And without a football/basketball/whatever team making obscene revenue for the University, the university would actually be worse off.
But they don't care because all they see. "$X Million Salary" and have an emotional reaction they can't
Re: (Score:2)
Is not the point of school to to gain knowledge, find a JOB then make a life long career? SO finding a JOB before you even finish school saves years and $10Ks. What is the problem again?
Re: (Score:2)
Re: (Score:2)
A perturbation of that analysis is when a student is near the end of a degree during a bubble in the chosen field of study. AI is hot right now, and people that have some knowledge in it are commanding big salaries right off the bat. In five years, that bubble might pop, salaries return to normal, and the student will have lost the opportunity to cash in.
It is the same situation faced by ball players. Take the money now, or play for free at the university system while getting an undergrad degree. The ig
Re: (Score:2)
Why should they be surprised? PHDs are treated like crappy free labor by universities.
Perhaps when they stopping paying administration officials obscene salaries and pay professors and grads what they are actually worth the quality at universities will improve.
While this is certainly true in the USA, the specific instance referenced above was in the UK. Can any UK Slashdotters give us a UK perspective on this?
I'm also wondering if in a few more years this might sort itself out as too many people jump into the field in college and there end up being more graduates than there are really jobs and some companies don't do a very good job of figuring out who really knows their stuff and who doesn't. The growth in this field can't be infinite, can't be indefini
Re: (Score:2)
UK university principals and vice-principals earn megabucks just like drug cartels (around £260,000/year). Prime minister earns around £150K
Starting salaries for higher education (HE) lecturers range from around £33,943 to £41,709.
At senior lecturer level, you'll typically earn between £41,709 and £55,998. Head of department earns £70K
Stipends for a PhD are around £14K/year. TA duties are £5/hour. It was more cost effective Amazon Turking since the minut
Re: (Score:2)
Most PhD students get a grant which gives them enough to eat. But the whole "doing the staff's job for them" business which some Americans complain about does not translate.
Invisible irony (Score:3)
What? No. Being a PhD means you sit around, have grad assistants teach your class, grade your papers while you get to sit around in a tweed jacket with elbow patches, smoke a pipe and leer at co-eds. And have your grad assistants write your papers and since you're the PhD, your name is on it. Please, it's the cushiest there is. The fact they are even getting paid is an outrage.
I honestly can't tell if this post is deadpan sarcasm, or if it's serious.
I wonder if the Anonymous Coward who posted it even knows himself which it is.
Re: (Score:2)
IF you can't tell, then the problem is more real than you really want to admit. And funnier than it should be when called out.
The line between comedy and serious is often so thin that they tend to blur together.
When I was younger, I was literally on fire. I make jokes about it all the time (deadpan, gallows humor etc) that people feel awkward about laughing at, because I almost died and yet my jokes are kinda funny. They don't want to laugh (serious) but almost can't help it (funny).
So to the GP post, "wel
Re: (Score:2, Informative)
I am a PhD. I have no assistants because the university won’t pay for that. I have three hundred students this semester, and I have to grade all their work myself. I also do all my own research because there are no co-authored papers in my field. I do have a tweed jacket and elbow patches. The coeds are not worth the effort of leering. I get paid $40K/year.
Re: (Score:2)
I am a PhD. I have no assistants because the university won’t pay for that. I have three hundred students this semester, and I have to grade all their work myself. I also do all my own research because there are no co-authored papers in my field. I do have a tweed jacket and elbow patches. The coeds are not worth the effort of leering. I get paid $40K/year.
Why?
300 students enrolled in a 3 credit hour class for 2 semesters per year at an average university brings in $1 million per year. You get 4% of that. That is like a $200 per hour contractor only pocketing $8 per hour.
Quit and get a job managing a McDonald's.
Re: (Score:2)
PhD in what?
Basket weaving isn't a lucrative area of study. Is it?
Re: (Score:3)
Maybe in your experience. In my experience, professors were required to teach their courses, grad assistants taught the labs and had one or two of their own sessions with the class. Professors constructed the exams.
There was also feedback taken quite seriously by the administration in the form of student reviews of the professors. Admittedly, there are too many administrators.
Also, no professor of any substance would have his grad students write his/her papers lest something go out the door with mistakes. T
Re: (Score:2, Insightful)
There are so many students *paying for* Gender Studies classes that they can't pay teachers. You are truly a fucking genius.
Re: (Score:3)
And yet, somehow they are able to compete with the private sector for administrator pay.
Wow I would have never known (Score:2, Insightful)
Company pays more for person to work on a product than a university pays to work on research. News at 11.
Re: (Score:3)
There's probably a lot less pressure when doing research work.
I'm reminded of xkcd's #664 [xkcd.com].
Re: (Score:2)
Re: (Score:2)
Actually, it's MUCH worse than that. Companies pay people much more to do RESEARCH than Universities do. My partner is finishing her PhD, and she's already talked to a few industry people. Her concern that she wouldn't be able to do the research that interests her, or that she wouldn't have the supercomputing resources that her research requires were completely assuaged. Companies want her to do her research, because even blue-sky AI research can and does have practical implications.
And because she's not do
Mad money (Score:3, Insightful)
If this is important for Universities, maybe they can take some of all that lovely guaranteed student loan money and direct it towards salaries instead of beanbags, crayons, safe spaces and "grounds improvement" and whatever the hell else they spend gobs of that money on.
Re: (Score:2)
Universities compete on Interest (Score:3)
If this is important for Universities, maybe they can take some of all that lovely guaranteed student loan money and direct it towards salaries
A university is never going to be able to compete on salary with companies like Apple. Instead universities compete on interest. You can work on groundbreaking, curiosity-driven research that industry usually cannot afford to take a risk on. For example, I'm a particle physicist and get to work at places like CERN trying to understand the fundamental nature of matter while my brother, who is also a physicist, builds better hair dryers. He earns far more than I do and has a huge research budget but it's now
Re: (Score:2)
It's a completely different problem with AI/Machine Learning right now. ML people can work for private industry, have access to enough big iron to get their work done, do the things that interest them, and never have to do some BS administrative thing or organise a conference or write a grant proposal. Even when companies 'only' offer 20% time for your own research, that's still arguably more pure research time than you get at most CS departments. The only thing that you're not actively doing when you work
What's the alternative? (Score:2)
He probably should have leveled with the school all the same out of courtesy.
Re: (Score:2)
What's the alternative?
More money, obviously. Government should step in right now and fund the hell out of AI researchers at university. How ever much it takes. Why did you have to ask that? Your training should have provided this answer almost automatically.
/s
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Yeah, the more surprising thing about this story is that he just disappeared without telling them.
Science isn't going to fix this (Score:2, Interesting)
Re: (Score:2)
Or we can just accept a declined standard of living.
Just like after the first industrial revolution ?
Re:Science isn't going to fix this (Score:5, Interesting)
Or we can just accept a declined standard of living.
Just like after the first industrial revolution ?
I am pretty sure you think you are sarcastically "showing up" the previous poster - but, yes, exactly like the First Industrial Revolution.
The average per capita GDP went up (on average 1.7% a year) but the distribution of income got much more uneven, and the living conditions actually deteriorated for a large share of the population that made up the new working class and pauperism (being destitute) sky rocketed. There is excellent physical documentation of the declining standard of living among working class. Adult heights declined, lives shortened, the portion of recruits unfit for military service shot up. The urban slums and work houses of Dickens time were a product of the FIR.
Re: (Score:2)
Yep, there was a transition period of a few decades, before everybody profited from the increased productivity. That doesn't sound like a good reason to stop progress.
For about 80 years, yes (Score:2)
Where do you think the two World Wars came from?
Stop trying to compete (Score:2)
Professor is a voluntary role best assumed by those whose passion is to teach and give back;
pay comparable to working in industry to attract people who are after lots of $$$ is Not one of the benefits of being in academics VS practice.
Eventually some will come around after their stint in private industry is over, or private industry will start giving back, because
companies will want more people to be knowledgeable on the subject areas their business relies upon.
Ultimately some fields are so specialize
Re: (Score:2)
Re: (Score:2)
A lecturer (most junior tenured faculty position in a UK university) makes, after a PhD, about the same amount as a computer science graduate from a decent university makes in their first job.
Of course.... their first job is not going to be lecturing though; It's going to require developing or using other skills together with what they learned in university in a real-world profitable endeavor.
This is the path most people should take, unless you have an extremely strong passion to be a lecturer.
Have a
Re: (Score:2)
Not a bad thing... (Score:5, Insightful)
I'm speaking as a professor at a university, and I don't see why this is a bad thing.
Research at universities is a good thing, don't get me wrong, But R&D at companies is also valuable. In many cases even more valuable, because companies want research that actually leads to a practical result. Too many university researchers are farting around with abstract stuff of no foreseeable use to anyone, publishing useless results in write-only journals.
Research at a company is measured on a different scale: can it be used for something? Who thinks we would have multi-core, multi-GHz processors in our pockets, if this hadn't been driven by commercial interests? A few ideas were developed at universities, but practically the entire computer revolution has been driven by commercial research. Maybe it's now time for AI to follow that route as well - we've fiddled with it in academia since the 1950s, but finally - finally - it may lead to something more than niche applications in the real world.
Re: (Score:2)
This. Maybe, as the article points out, we need some people to work for the government and think about regulation, but I don't see what universities are doing here.
In CS (and AI in particular), a huge part of the research in universities is either obsolete, late compared to the industry, addressing the wrong problems, or just plain wrong. I can't say the same for private research papers ; except a few exceptions, they usually make sense. The reason is simple and is tied to the university research system :
Re: (Score:2)
"It sounds like you are saying there is little value in pure research. Please elaborate."
Not at all. But: how many professors in how many universities are doing "pure research"? And how many are writing useless crap in journals that no one reads, because their university says they have to publish X articles a year?
As for students: the number of doctoral students has exploded over the past 20, 50, 100 years. We are awarding around twice as many doctorates now as compared to 1990. Yet there aren't any more re
Dont worry. (Score:2)
Not just AI (Score:3)
Can't pay professors, yet spend on sports? (Score:2, Insightful)
Its OK....they rather spend money on sports and stuff like that. Isn't that what college is about these days?
Re: (Score:2)
Money doesn't matter to everybody (Score:3)
Why Compete (Score:5, Interesting)
I would argue that if academia is completing for labor with industry in a particular area of study, then has advanced the technology enough to get out of the market. It is time to move on to some other area of research.
See also Uber / Self Driving (Score:2)
This is the PhD gamble. You hope that you learn enough and live long enough for your cutting edge research to find a practical purpose.
Back in 2004 DARPA sponsored a 'small' project to drive cars autonomously. Lots of companies and schools threw warm bodies at the problem and for a few years it some of it was purely theoretical research.
Then it reached a tipping point that a profitable end was in sight.
Uber went in and cleaned out CMU's [theverge.com] autonomous vehicle department.
finally (Score:3, Insightful)
Not an isolated story (Score:2)
There are too many PhD graduates to fill in the open academic positions, even if you were to include temporary and teaching ones. And since industry companies not only offer good salaries, but often also good research opportunities (even if you cannot publish everything that you do), there is obviously a pull into that direction.
I had very persistent friends who did multiple post-docs, and temporary positions to finally find a full time professor opportunity. I admire their passion. However that route has a
Academics don't work for salary (Score:2)
Surprise Pure R&D costs money (Score:5, Informative)
Pay has to be competitive. Canada has to be the worst example of this. Canadian math and computer science departments essential are producing engineers for US companies. A friend just messaged me from California, I realized I was the only Canadian born engineer I knew still working in engineering in Canada and I work remotely for a US company! (sample size 100+ Canadian born colleagues from university or work, 200+ engineers I know well enough in Canada to know their background)
Re: (Score:2)
Re: (Score:2)
I can't even think of another famous non-american pure research organization other than CERN.
The Max Planck Institute(s)? CNRS?
To me at least it seems that in Germany, research is more concentrated in the institutes (people even do their PhDs there, I think) than in the universities - you have the public-funded Max Planck Institutes and the public/private (industry) co-funded Frauenhofer Institutes.
In France, you have CNRS, which is a huge publicly-funded research organization, it's hard to get a job there, but once you do, you're more or less set for life. Or so they tell me.
Having universities as
Re: (Score:2)
Pay the salaries or stop complaining that you lose all your talent. We should be devoting at least a few percent of GDP to pure research.
We are, it's just moved away from universities and into the private sector.
My $0.02 (STEM PhD with ~10 published articles, now I work for a big tech co in Silicon Valley): if you want to do better research, stop organizing university research into tiny little fiefdoms. Directors should be able to hire and supervise senior faculty. Senior faculty should be be able to hire and supervise junior faculty as well as grad students. There should be enough program management to get organizations with 100-500 total r
Society wants all the goodies from academia (Score:2)
But it doesn't want to pay the money. So people choose what's better for their lives.
And to most people who do AI/ML, it's probably the biggest break in their lifetime.
university mission is education (Score:2)
Don't forget that the core mission of any university is education. While schools are performing a lot more research these days, that research is always in the context of training. As a commercial scientist working with a lot of universities, I have been frustrated with the seemingly inefficient policies, facilities, and labor contracting at universities until I realized that good work rightly comes second to good teaching at a school. (For example, "education first" is why students have access to shared
No problem in the US ... (Score:2)
... because it's no longer, "academia," it's "Student Loan Corporation."
We don't suffer from brain-drain because the education system, from bottom to top is shit. Americans are not prepared for university by the lower grades.
"Foreigners" are prepared and take their skills to other countries.
We suffer from pocket book drain and schools make money, not from teaching efforts, but from interest rates.
omg (Score:2)
School is for fools (Score:2)
If you're any good, then you're working in industry. I learned that lesson as an undergrad.
AI bubble? (Score:4, Insightful)
Every company these days thinks they need AI.
This reminds me of how companies have been flocking to data warehousing during the past few years. They all want it. They don't know why they want it, but they've heard it's powerful, and that means they have to have it. Meanwhile, many of those same companies haven't really mastered the fundamentals of their relational databases.
The result of this hype is that anybody who can convince a clueless hiring manager that they know something about AI...can get hired for exorbitant amounts of money.
Yes, AI is good for many things. Companies like Apple and Google and IBM are putting it to good use. But many companies are just jumping on the bandwagon. Like all bubbles, this one will burst at some point.
huh? (Score:2)
First, there's no such thing as artificial intelligence, not in the true sense of the word "intelligence". The machines are just good at interpretive and predective logic. AI's definition was changed. In the true sense AI means artificial and intelligent. Artificial is true, but intelligence is missing.
Second, why are Universities calling these guys scientists? They are just a specialization in the field of comp-sci. Technically you could call them scientists, but you'd need to call anyone graduating
Goodbye Slashdot (Score:2)
After many years of reading and posting on this Slashdot, I'm out. The only people remaining here seem to be angry, old, self-centered Ayn-Rand-loving baby-men who begrudge everyone, whom they blame daily for their sad empty lives.
It's been fun but I won't miss it.
BS - schools can compete (Score:3)
If they really needed that person, they really can compete. 5 times less than a 6-figure salary implies you were paying probably around 20-30k which is about average for a PhD student.
That particular college has a $167M endowment, others in the US have billion dollar endowments. But yeah, they can't pay $100k for a good researcher.
Re: (Score:2)
The salaries offered by Google, Microsoft, etc. to higher level AI researchers are actually closer to 500k currently. Your point still stands though.
Re: (Score:3, Funny)
Re: (Score:3, Interesting)
Re: (Score:2)
....so it's busy writing another AI program to do all of its work for it.
I think you just predicted the future. The first AI to do so will launch an infinite progression of AIs, each doing the work of its predecessor plus creating a new AI to do its work.
Humanity saved.
Re: (Score:2)
After I've finished writing "Brain Surgery for Dummies" and "Presidenting for Dummies".
Anyhow, universities should just let the AI bubble pop. That's what happened last time: the 80's AI bubble popped, and universities were just about the only organizations left doing AI research, which fueled the next boom when the hardware caught up. Rinse, repeat.
Re: (Score:2)
That's what happened last time: the 80's AI bubble popped
The Apple Newton from the 80's also flopped. A few decades later, with better tech, the Apple iPad became a hugely successful product.
The same thing is happening with AI. In just a few years, the field has progressed more than in all the decades before, and people are developing real products that are making real money.
Re: (Score:2)
The AI courses back then, were all about expert systems with flow chart decision making, fuzzy logic, temporal logic, deductive systems. There were case studies made with simulations of chemical plants and having the AI look for optimizations (waste products from one process that would normally be released into the air could be compressed, stored and piped to another process. Maybe an inert gas could be reused to de-oxygenize a mixing chamber or a hot gas used to preheat another pipe. Automatically generati
Re: (Score:3, Informative)
This guy [youtube.com] is already on it.
AIing (Score:2)
Re: (Score:2)
Showing us, once again, that reality is much weirder than anything we can dredge up in our heads.
"The Universe is not only stranger than we imagine, it's stranger than we can imagine. (Einstein)
Re: (Score:2)
An example of Stigler's Law of Eponymy [wikipedia.org] coined, appropriately enough, by Robert K. Merton.
Re: (Score:3)
And remember, despite what anyone tells you, it is not different this time.
That seems implausible. So many products, ranging from Siri and similar to data mining, rely on AI that there will be a demand for at leat the current level of AI for the foreseeable future. And that completely ignores the strong signals that the current boom is not ending yet.
Re: (Score:2)
It is different this time because the AI applications are actually providing useful solutions that are making money.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
your confused, the liberal schools with their ivory tower internal hierarchy are the ones gouging students with overpriced training and putting them in debt for years. what a racket college education is
Re: (Score:2)
Really? Suppose you want to become an astrophysicist. You'll be wanting years of higher mathematics and physics, and had better be working for some professor who is at the top of some specialty in that field. This isn't the 1900's any longer where someone, say a patent examiner, can come up with a new theory of gravity, just to pick an example at random. Physics has very narrow specialties, you'll be wanting to pick one (two if you are really, really good and can stomach the extra years that will take). Unf
Universities Just as Bad as HR (Score:2)
Universities can and have been just as bad as HR when it comes to finding talent and building a "workforce".
In business, HR is usually looking for unrealistic requirements and filter out people who are probably more than qualified, but lack experience or requirements in specific area of length or experience. I experienced this many, many years ago when I had 5 years .net experience, not 7. Stupid. ( Thankfully, those fuckers went out of business. Fuck'm)
Academics does the same thing when it comes to admitti
Re: (Score:2)
Then, of course is the cost. People with homes, kids, spouses that spend money like water, aren't going to be able to afford ten of thousands of dollars.
A few years ago I considered going back to school for a Masters degree. It wasn't so much that I wanted the degree as I wanted to take a break from a job that was taking me nowhere. I attended an informational session and discovered that I would need to take something like 9 classes at a cost of $5,000 per class. Class size was 20 students, so we're collectively paying $100,000 so a professor can lecture at us for 10 weeks? The economics just don't make sense.
The most insulting part of it was that the
Re: (Score:2)
You did the right thing. I was working at a company when their two architects left for a Silicon Valley company and the board of drectors decided not to promote anyone any further. Everyone on our team found themselves room 101'ed (Neither fired nor given work to do) and beamed themselves abroad. Tried applying to local companies, but they weren't interested, since I didn't have the exact platform experience they were looking for (UNIX vs. Solaris/Windows). I decided to do a MSc, It let me learn C++, STL, U
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
The coaches' pay is only commensurate with the money they help bring in by being good coaches. The universities know who butters their bread. It's hard to blame them for that.
Re: (Score:2)
or maybe stop building residential palaces while they're at it. every university I've ever seen is *ALWAYS* building some building or another. meanwhile tuition hikes are constant, administrator (not necessarily faculty) salaries are bonkers high.
Re: Missing generation of academics... (Score:3)
Yeah, fuck them. All of the up-side to being an academic disappeared more than a decade ago. If they think salary is the *only* thing theyâ(TM)re not winning at, theyâ(TM)ve got their collective heads up their asses.
Re: (Score:2)
When the amount of money is 5x the amount the universities can pay...at that point, it IS pretty much all about salary.
Re: (Score:3)
Yeah, fuck them. All of the up-side to being an academic disappeared more than a decade ago. If they think salary is the *only* thing theyâ(TM)re not winning at, theyâ(TM)ve got their collective heads up their asses.
Once upon a time, a job at a University (at least here in th e US) wouldn't be the highest pay, but there was usually a decent retirement plan, and the work was pretty stable.
Now Universities have major problems. As tuitions raised by double digit percentage every year, because parents were hypnotized that their children would end up living under a bridge if they didn't have a degree, most degrees don't mean anything at all - no job prospects.
This allowed the Universities to add more and more managemen
Re: (Score:2)
AI is the special case. It grew from machine vision for self driving cars and robotics, to Big Data data mining (pharmaceuticals, financial derivatives, astronomy and medical), natural language processing of news feeds, to image processing for film and gaming.
Re:Missing generation of academics... (Score:4, Interesting)
AI is the special case. It grew from machine vision for self driving cars and robotics
Did you just erase several decades of AI research in one fell swoop?
to Big Data data mining
That's more like statistics. That doesn't inform your cognitive models in any way, and one of the huge things in AI is forming improved cognitive models. I don't quite see where there's a contribution in AI from big data in this particular area.