Slashdot Mirror
Senate Passes 'No Microsoft National Talent Strategy Goal Left Behind Act'
theodp writes: Microsoft is applauding the Senate's passage of the Every Child Achieves Act, a rewrite of the No Child Left Behind Act, saying the move will improve access to K-12 STEM learning nationwide. The legislation elevates Computer Science to a "core academic subject", opening the door to a number of funding opportunities. The major overhaul of the U.S. K-12 education system, adds Microsoft on the Issues, also "advances some of the goals outlined in Microsoft's National Talent Strategy," its "two-pronged" plan to increase K-12 CS education and tech immigration. Perhaps Microsoft is tackling the latter goal in under-the-radar White House visits with the leaders of Mark Zuckerberg's FWD.us PAC, like this one, attended by Microsoft's William "It's Our Way Or the Canadian Highway" Kamela and FWD.us President Joe "Save Us From Just-Sort-of-OK US Workers" Green.
Everything I've read about deep learning is that they are great pattern recognizers. However, they are no closer to coding than they are to writing novels. What they are exceedingly great at is taking huge volumes of information and finding patterns. So the jobs that should be concerned are stuff like X-Ray technician, or general medical practiioners whose job is essentially managing a lot of data that most people haven't had the time to memorize/learn. The most disrupted job, will be truck drivers (the number 1 job in many states). And yes, driving is still just pattern recognition - stop at red lights, don't hit the car, when a car does this, you do this - type stuff. Give it a large enough amount of data and it can figure out what the results should be.
Once you start getting into creative pursuits, deep learning is no closer to that than it has ever been. It might be able to elevate patterns that we previously didn't realize were connected, but that's about the limit of it's "creativity".
What you won't get with deep learning (so far), is saying "take this API, then take this other API, then produce this unique solution that I have come up with for my unique problem". Example: We need a UI for a gas pump that will handle new EMV cards - deep learning, write a solution. Not happening. So I can only assume a statement of "coding will be obsolete in a few years" is either hyperbole or ignorance of the state of the art.
1. There are already more STEM graduates than jobs.
No. STEM fields have an unemployment rate of about 3%, compared to about 5% overall.
3/4 of STEM workers leave the field due to poor pay and working conditions compared to other jobs.
Nonsense. About 75% of ALL college grads work outside their major. STEM majors are more likely to work in their major, and those that don't frequently work in other STEM fields, such as physics majors working as programmers.