Google AI Education Resources | Hacker News Reader

Google AI Education Resources

ai.google

300 pointskilljoywashere31 comments

I really like the Introduction to Machine Learning Course that Google has put up on their developer site [0]. It's simple enough that anyone can get into it, while still digging deep enough into the math behind things that it isn't just you learning how to copy and paste code.

Besides reading an actual textbook or learning from an online course like (https://eecs189.org/) which can understandably be somewhat daunting at first, I think this is definitely the next best thing.

[0] https://developers.google.com/machine-learning/crash-course/

I think it's really cool that Google is so motivated about creating and releasing educational content that anyone can use to become a great developer. Their tech dev guide is also pretty helpful (although in fairness, their cloud education does focus on GCP pretty heavily).

Google's play seems to be to encourage more and more folks to learn CS and be interested in learning programming and AI. Google Research, GSoC, PhD research grants, publishing their results in leading journals, and their OSS contributions all point to a company that is a friend of programmers and academics. For these very reasons, the company is viewed very positively outside of SV.

On the contrary, I don't see a similar desire to contribute back from Amazon, Apple, Facebook or Netflix (although the last 2 do contribute a ton to OSS).

Most of googles evangelization of tech and AI is a marketing and adoption strategy for google compute engine. Which is fine, but it should be seen for what it is

Disclaimer: I am personally benefiting from a long-term Google educational programme.

My response to this is: so what? Of course they're going to encourage new developers to use their products - it's only natural they'd teach their own stuff and not AWS/Azure. Developers aren't really influencers (sadly) in the purchasing for cloud infra so I'm not sure that it's a straight up marketing campaign.

Google was publishing very important distributed computing papers and other research even when they didn’t have cloud offering in mind.

It’t a great hiring strategy, as many researchers prefer to work at a company where they can get recognition from publishing academic papers.

30 seconds googling yields....Amazon (https://aws.amazon.com/blogs/machine-learning/amazons-own-ma...), Apple (https://opensource.apple.com/, https://github.com/apple/turicreate)

Here is this free course that grants a certificate for the University of Helsinki, https://www.elementsofai.com/ it is for beginners, some banks are using it for their staff too.

For me the biggest problem, as manager to build AI team, is everyone talks jargons of AI/DL through these kind of courses. How do I find out who knows in depth? Do you have any intriguing questions you have that can quickly separate the wheat from chaff? Is there any Questionnaire on AI/DL that help select (AI expert)/prepare (for Job interview)?

Same as you'd select any other professional where you don't know how to do their job.

1) Work experience. 2) Educational pedigree.

There's also the idea you should hire a program lead who does have demonstrated experience, and let them do the technical interviews of their underlings...

At the companies I've worked for doing scientific programming work, during my interview I had to give a presentation to the engineering team over lunch concerning a technical problem and take questions. If a candidate was a fresh graduate we had a requirement they have 1) a Master's and 2) they would present on any research papers they had published, UNLESS 3) they had interned at our company and everyone liked them, then they could skip the Master's requirement.

> research papers they had published

I wonder how academic publishers feel about this growing trend to require not just a degree, but now peer-reviewed, published, research papers to get a job. The sentiment behind it is, of course, if you’re really smart, you would be published, so if we look for people who are published, we know they’re smart. The reality, though, is that once this catches on, journals are going to be flooded (even more than they already are) with desperate attempts to get something, anything, with somebody’s name on it since that’s another checkbox they have to tick before they can eat. Just like what happened with higher education.

Ask them about statistics and when it's appropriate to use certain models over others. It's basic, but pretty revealing if a candidate doesn't understand basic tradeoffs between bias and variance when it comes to model selection.

Knowing about the bias-variance tradeoff is like asking a candidate to pass Fizzbuzz. It doesn't tell you anything except that they aren't completely fraudulent.

Going into more detail isn't something that a non-technical manager type would be able to do.

As a non-technical manager it will be very difficult - just make sure that the selection process has a technical component. As a technical manager (and opinions here will vary), focusing on linear algebra is a great filter, because it's essential to anything ML but it's rarely taught in most online courses.

this might be relevant for you. https://www.deeplearning.ai/ai-for-everyone/

I'm a backend developer who uses Java an GoLang regularly to write back end systems - APIs, workflows, infrastructure, deployment pipelines, etc.

I don't have any knowledge about AI or how to model problems for recommendation systems, or when to use decision trees, versus something else. Is this a skill that I should be actively investing in to not become a dinosaur?

My worry is that in the next 10 or so years, I don't want to end up as a Cobol developer in the world of today i.e., might have a job and good pay, but not being able to work at the next big company or next big idea. What are your thoughts?

When databases came by, I bet people felt the same way about them. Now they're a common tool used in most systems, and most devs and engineers will have at least a basic understanding.

My hypothesis is that ML will follow a similar path. It seems like an exotic skill now, but there's already a mass of undergrads familiar with it from their education. We'll still need experienced practitioners to lead projects and architect systems (like DB admins and architects do!). But in some 10 to 20 years, everyone will use ML where appropriate, get some value from it, and it will have lost its hype. There will be some uniquely new capabilities that ML enables, just like DBs enabled storing state at scale, efficiently and cheaply.

Indeed. ML won't be the be-all end-all of software engineering, but it'll be a core skill to have significant knowledge about. Even if you're not developing new models, knowing the best practices for using them or extending them is going to be invaluable.

I'm in a similar position. I did a bunch of courses on coursera, udemy, pluralsight, fast.ai -- because I like ML stuff, however, I'm pretty sure this will be just a hobby.

I've come to the conclusion that AI/ML is a winner takes all market, I really don't think we'll see millions of ML/AI "developers" like we've seen C++/php/.NET/Java in the past.

For the next 10 years I'll focus on kuberbetes/some big data/some ML, but even though I regret it, I think the only viable option is to get into management.

Learn as much about ML as you're interested in but don't worry about becoming a dinosaur at all.

Nah. It's a specialized skill that most programmers won't need to know. Is it useful? Yes, but only if you're a specialist, just like any other niche in computer science. Nobody needs an amateur machine learning engineer, just like nobody needs a hobby compilers enthusiast on their team, or weekend warrior cryptanalysis abilities.

Be an expert or go home.

I work as a Mechatronics engineer and I have an interest in AI. I've personally gone through a lot of the online resources out there: 1. Andrew Ngs Deep learning MOOC

2. Fast AI parts 1 & 2

3. The old Google Machine learning course

But, what next?. From my experience, this doesn't give you enough credibility to get you a job interview at even a small sized firm, let alone Google.

Don't get me wrong, I really appreciate all the fantastic AI learning resources out there. Its incredibly enabling, but I feel like I'm missing the point of this - Is it to enable people to start companies using AI based tech, and grow the google compute based ecosystem? If its to grow the number of AI jobs and eligible people for those jobs, I have doubts whether that's actually working, or am I missing something?

I'm not an expert, but I don't see academic courses here such as e.g. [1]. And I don't see books read such as [2]. Personally, I would follow those type of things as grads or undergrads would follow the same courses, and on top of that I'd do what you do.

[1] http://cs231n.stanford.edu/

[2] http://neuralnetworksanddeeplearning.com

I'm in Ngs course now and also half way through fast.ai. I'm also interested to learn whether these courses reliably lead to anything. I gather from Jeremy's comments in fast.ai that those who attend in person reliably get good jobs in the field. So networking appears to be the key. I wonder what can be done to improve the networking opportunities for people who take the online courses?

Surely the easy answer is do something with your knowledge. If you feel you can apply it, then I would say it was useful.

I couldn't imagine reading 3 books on python, and wondering will I get an interview. The question should be, can I write a simple program. Measuring by can I get a job interview is asking the reverse question.

I mean, would you hire you? Can you solve a potential company's problems with your AI toolset.

I get what you mean. I've been applying the skillset to Kaggle problems, each of which I imagine contain multiple subproblems which companies might face. But kaggle standings, in my experience, dont seem to be too convincing a metric for job openings.

The problem with the MOOC ecosystem at the moment is there's no clear path forward with them. I'd have imagined the MOOC certifications solving this problem, but I feel networking plays a much bigger role in the job market rather than credibility.

The only exception I see is Udacity, which, by its pricing has created a limited pool of graduates, and therefore are valued much higher

Stay away from:

- MOOCs

- Udacity

- Kaggle

I'm not being facetious, this is my honest advice.

I am under the impression that the courses are designed for EE/CS engineers to get familiar with the foundations of modern ML, but it's not sufficient education to work as a full time ML engineer.

I returned to grad school for ML two years ago, and even now I still struggle with some ML job interviews when it comes to statistics and theoretical questions that I've studied two years for. One particularly challenging part of ML interview is that it covers much more than a typical CS interviews that I'm used to. I had a ML engineer internship interview with a famous ML company recently, and I was asked about sorting algorithms, hashing algorithms, non-convex optimization techniques, gaussian processes and manually compute the jacobian of a NN for backprop on the spot.

    > but I feel like I'm missing the point of this

There is no overarching point - each of the resources that you've listed have their own reasons to share educational content for free, it's up to you to use it as well as you can.

Well, you're right. The people sharing the resources are doing it for education's sake - for anyone that's interested. I think it's more accurate to say I feel locked out

There’s a misconception out there about the data science skills gap - the truth is there is a huge demand for highly skilled data scientists, a big demand for data and ml literate developers, and a moderate demand for entry level data scientists.

These resources from google and courses like Fast AI are great for getting devs up to speed so they can meaningfully contribute to data science projects - filling that big demand for data + ml literate devs, especially internally. They’re not designed to get people jobs (disclosure, getting people jobs in data science is what we do at thisismetis.com)

If you want to go deeper? The open source data science masters is a good set of resources[0]. The first few sections of Goodfellow’s deep learning book are a great crash course in ML math/stats theory[1]. Introduction to Statistical Learning is a staple in most people’s library[2]. There’s a glut of intro level data science content out there on the internet, but intermediate to advanced stuff usually means putting in serious effort or breaking out your checkbook and going back to school (whether traditional or otherwise).

[0]http://datasciencemasters.org/ [1]https://www.deeplearningbook.org/ [2]http://faculty.marshall.usc.edu/gareth-james/ISL/

Reply on news.ycombinator.com