Elon Musk Wants A.I. Developers, No Degree Required

Tesla CEO Elon Musk wants to hire some artificial intelligence (A.I.) researchers—and he says he doesn’t care if they actually have degrees.

Earlier this week, Musk Tweeted: “Join AI at Tesla! It reports directly to me & we meet/email/text almost every day. My actions, not just words, show how critically I view (benign) AI.” The accompanying link led to a page for Tesla’s autopilot projects, which include a combination of hardware (chips that power A.I. software) and software (neural networks, autonomy algorithms, and more) initiatives.

In a subsequent Tweet, he added: “Educational background is irrelevant, but all must pass hardcore coding test.” That message also included a list of the programming languages utilized at Tesla, including Python:

In a response to another Twitter user, Musk clarified that educational aspect. “A PhD is definitely not required,” he wrote. “All that matters is a deep understanding of AI & ability to implement NNs in a way that is actually useful (latter point is what’s truly hard). Don’t care if you even graduated high school.”

Musk is infamous for maintaining a workaholic schedule as the CEO of two companies (in addition to Tesla, he also heads SpaceX), and so a job that requires constant communication with him might not suit all employees, unless you really enjoy a constant stream of emails and phone calls at 3 A.M.

That aside, Musk and Tesla seem to be following a broader trend in the tech industry. The heads of IBM, Apple, and other companies have taken pains in public interviews to emphasize that they often hire for skills, not educational pedigree. “We’ve looked at the — sort of, the mismatch between the skills that are coming out of colleges and what the skills are that we believe we need in the future,” Apple CEO Tim Cook told the audience at an American Workforce Policy Advisory Board Meeting in early 2019, “we’ve identified coding as a very key one.”

Even if you don’t need a degree to land a gig, though, you absolutely need the right mix of skills. Fortunately, when it comes to autonomous driving, there are some courses online that can introduce you to the skill-sets necessary. These include Udacity’s a self-driving car engineer nanodegree (Udemy and Coursera have similar coursework). For a very deep dive, MIT has videos and slides from its “Deep Learning for Self-Driving Cars” coursework available for your perusal. 

And of course, no autonomous-driving education is complete without learning everything possible about artificial intelligence (A.I.), machine learning, and computer vision. If you want a solid intro to machine learning, Google has a (free) course with 25 lessons and more than 40 exercises, which you can finish in roughly 15 hours (and features lots of interesting video). For those who already have some knowledge of machine learning, and deep familiarity with math, there’s also Bloomberg’s Foundations of Machine Learning.

If you actually apply for a job as an autonomous-driving engineer at Tesla, you’ll have to endure that intense testing regimen that Musk mentioned in his Tweet. Before heading into any tech-related job interview, it’s always a good practice to brush up on your coding skills.

8 Responses to “Elon Musk Wants A.I. Developers, No Degree Required”

  1. No wonder Tesla’s A.I. “self driving” programs are full of glitches and killing people! Can’t wait to see the carnage from the space program! I think Tesla’s should be required to have rooftop warning beacons for other motorists and pedestrians that flash anytime this flawed technology is being utilized!

    • Any new technology that comes out for testing and installation for ordinary use is probably going to have some glitches involved in it’s infrastructure. Humans are not perfect and probably, some mistakes are made and corrected but, there will always be something missed through the cracks of achieving perfection. To make a perfect autonomous vehicle, one will have perfection only after numerous attempts and failures. Sometimes it takes more than one generation to create perfection in such complex technologies. You have to ask, is the technology available to make perfect autonomous vehicles?

    • ARLibertarian

      Nothing is fool-proof. But self driving is likely better than many human drivers, and destined to improve. On your commute, look around the the people fiddling with their phones, driving aggressively, eating, smoking, chatting, texting. How many wrecks a day do fallible humans cause? If self driving has a lower failure rate than humans, then it’s progress.

    • Paul Smith

      Not requiring a degree simply means you don’t get the job only because you have 8 or 10 years of college, there are lots of very talented people that know coding frontwards and backwards that are self-taught or learned with little schooling that can code better than those with a college pedigree. He is not limiting his team! That is the take away.

      • Mike Piazza

        I’ve worked with both types of coders and I’ve also worked with coders that had other types of degrees, i.e. non-Engineering degrees and it really is a toss up. But being clever at coding only goes so far. If no one else can easily understand what you’ve written then your code is CRAP and not maintainable. That, I’m afraid, is all too common and has been for a long time irrespective of whether the person went to school or was self-taught. Creating maintainable code is not taught in University as the only thing you are trying to do is learn the course concepts and get your programs to work. Unless you are extremely disciplined when teaching yourself and care whether you understand something you wrote 6 months ago or not, learning to code by hacking your way thru what works and what doesn’t isn’t going to give you any more ability to write maintainable code either. In my 35+ years writing and maintaining software the absolute worst offenders have been C/C++ programmers. When I was starting out there was the obfuscated C contest, ffs, where it was a badge of honor to be able to write the most obtuse non-understandable code that actually did something. All that succeeded in doing was teaching people that it was somehow OK to write production code just like that. As a result a lot of crap code was written and I’ve lost count of the number of codebases I’ve had to try to figure out so maybe I’d have some chance to figure out what they were trying to do. What I came to realize was that it was dangerous to have a compiler like CC that is as happy to compile Procedural C code as it is to compile Object Oriented C++ code. Up to now, without management buyin there really hasn’t been anything to enforce it one way or the other and in the ‘early’ days of C++ it was a real problem as Procedural programmers were having to make the transition to Object Orientation and many heavily cheated just to get things done. However….nowadays most come out of school or training with a foundation in Object Oriented coding are no longer having to cheat by using Procedural shortcuts and with the coming emphasis on Dev Ops and the CI/CD pipelines with management’s enforcement of automated tools and testing suites there just might be a light at the end of the tunnel.

        • ARLibertarian

          Exactly right. Put these guys in the oncall rotation, let them get woken up at 2am to fix a critical job that’s down and has to finish before systems can come up at 7am. Maintainable code is 10x as important as showing everyone how clever you are.