a few weeks Ago I met someone, the manufacturer is at a large German automobile for the topic of Autonomous Driving to be responsible. I asked him whether his company has difficulties to recruit developers for machine Learning.
The first response of the car Manager was surprisingly upbeat: no, it's not as difficult as it was. Then he pushed that you have introduced in his company a new career path: A skilled developer can get as a starting salary in the future, more than the supervisor of his supervisor deserves.
300,000 to $ 500,000 per year starting salary
It is easy to guess how this amazing Arrangement is Different, you get easy to anyone.
The "New York Times" reported soon as 2017, that KI-specialists in the USA could expect starting salaries of between 300,000 and 500,000 dollars. For comparison, The top earners among the German professors get a little more than 100,000 Euro salary per year.
Here are some Numbers, and you will looks amazed
As for us, in terms of machine Learning really can be read in the Conference papers at major international conferences. Among computer scientists by Peer Review selected contributions to the conference are Central for the reputation.
The main AI conference in the United States is NIPS stands for Neural Information Processing system. The NIPS 2017 3240 Papers were submitted, of which 679 were accepted. 91 came from Google and its subsidiary deep mind, a further 40 from authors who work for Microsoft. Among the 57 institutions that took more than five Paper at the meeting, five European and not a German.
But perhaps at a European conference?
This is not to say that German researchers were not represented there at all - but there were very few.
It is good that German Chancellor Angela Merkel, "Germany and Europe in the area of Artificial intelligence, the worldwide leader" wants to make. At the Moment, that sounds about the same as the promise of Greece to win the next football world Cup.
Quiz digitization What is Crystal without Meth?takes place Well, the NIPS in California and travel budgets are always too small. Maybe German researchers submit their KI-a Work better at a major European conference? As the International Conference on Machine Learning (ICML), held in 2018 in Stockholm?
13 percent of the ICML Paper come from Google
58 institutions there were more than five accepted papers, and among them two German: Max-Planck-Institute for intelligent systems and the University of Tübingen are all the same. Together, they come to 21 posts. That is remarkable - just as much as the total number of successful submissions from Facebook employees.
if you Count together all of the Google Paper at the meeting, 82. That's 13 percent of all successful institutions.
Parallel to what Superbahis is going on in the USA, is happening in terms of KI, of course, in China very much, the Chinese developers are not only on Western conferences-so-present. Maybe it's because you keep your results to yourself.
80 percent to Google and Facebook?
Not that here misunderstanding: It is all well and good that American corporations share at least some of their findings with the scientific community. For democratic societies, but it is sometime problematic, if the Expertise for a centralized, almost all areas of life in question, rapidly world-changing technology of the future will be privatized.
A recruitment specialist U.S. entrepreneur has in the beginning of the year in "Forbes" claims that 80 percent of all U.S. high-school graduates with a relevant PhD ended up currently at Google or Facebook.
I presented these Figures in a recent lecture at the Paderborn Nixdorf-Symposium, in the secret hope that the computer scientists would speak in the audience to me. Unfortunately, that was not the case. It's just depressed gave lubricated Silence.
The three magic ingredients of the AI-research
to make in machine Learning, fast progress, you need three things:
1 computing power - the is when Google and co. as we know, in Abundance, and specially for machine Learning developed special chips. The exponential growth of AI invested in computing power is largely driven by Google.
2 Large amounts of data and is also available in the Silicon Valley as we know. Did you know, for example, that you create every Time you identify a Captcha, stones or Zebra stripes Schorn, data for Machine Learning? And for Google?
3 innovation , so smart people. The work, of course, also preferably, where computing power and training data en mass. And where you can earn five to ten times as much as a Postdoc at a German University.
You can blame the Federal government that they would not have recognized all of this. You want to allow "internationally attractive and competitive working conditions," the AI strategy paper. How to go to six years stretched to three billion euros, is beyond me. For comparison: For investment in transport infrastructure 14 billion were available only for 2018.
More on the topic of AI-plans of the Federal government, we Hear an Echo?100 new professors in the AI area are a good idea, because maybe the one or other Tip would like to research computer scientist rather than to increase the Power of the digital oligopoly. But that is not enough.
universities should make it easier for your top executives to earn their inventions. You would have to go out and offer better working conditions - and especially young researchers, perspectives go, the win over "those who get a chair,, all the others lose"lottery. This is significantly more expensive than three billion. But with money from latecomers to the "worldwide leader", will hardly work.
Date Of Update: 18 November 2018, 20:00
