Is age 40 too old to enter the AI/Machine Learning field

[FallenApple]

Ok, some background. I'm nearing age 30. I have a degree in applied mathematics and masters in statistics. I'm thinking about entering the artificial intelligence field( maybe computer vision) by getting a PhD in Computer Science.

It's not like I'm starting from complete scratch as I have a background in data analysis and programming. However, I really want to become an expert in leading edge AI tech. So this requires that I do the prep work to be qualified to enter a PhD program. By the time I complete the prerequisites would already be about 2-3 years. A second bachleors if need be. Then I would have to finish the PhD on top of that. So being conservative, I would say late 30s early 40s is when I can enter industry.

At age 40, how far off would I be compared to my mental peak? Far enough to have a real world negative effect?

I would need to be able to hit the ground running at that age.

 

[jrmichler]

You are as old as you let yourself be. I'm 66, just received my most recent patent last year, and have two more active patent applications. I can't have mentally deteriorated too much. Physically is a different matter...

 

[StatGuy2000]

@FallenApple, since you already have a masters in statistics, why not consider pursuing a PhD in statistics? There is considerable overlap between statistics and machine learning -- in fact, within my alma mater, there are 2 active professors (and 1 just recently retired professor) who were cross-listed in both the statistics and CS departments. Plus you have people like Rob Tibshirani and Trevor Hastie at Stanford, or Mike Jordan and Peter Bartlett at Berkeley who are involved in research at the intersection of statistics and machine learning/AI -- they accept students from both their own statistics or CS departments, and they are far from being the only places that do so.

A PhD in statistics for someone like yourself may well require fewer prerequisites, and you are likely to complete the program more quickly than a PhD in computer science.

 

[StatGuy2000]

And to address your specific concern about being too old for AI: one of my professors at my alma mater (the one I mentioned who has just recently retired) finished his PhD in Computer science when he was nearly 40, after working for years in industry. And he became a world-renowned researcher in machine learning/AI and statistics.

http://www.cs.toronto.edu/~radford/cv.pdf

 

[FallenApple]

@FallenApple, since you already have a masters in statistics, why not consider pursuing a PhD in statistics? There is considerable overlap between statistics and machine learning -- in fact, within my alma mater, there are 2 active professors (and 1 just recently retired professor) who was cross-listed in both the statistics and CS departments. Plus you have people like Rob Tibshirani and Trevor Hastie at Stanford, or Mike Jordan and Peter Bartlett at Berkeley who are involved in research at the intersection of statistics and machine learning/AI -- they accept students from both their own statistics or CS departments, and they are far from being the only places that do so.

A PhD in statistics for someone like yourself may well require fewer prerequisites, and you are likely to complete the program more quickly than a PhD in computer science.

Ah ok. That makes sense. So it would make more sense for me to do the Stats Phd. I guess I would have to take CS courses on the side as well. Because I think I would need to specialize in a topic in order to properly do artificial intelligence. Such as computer vision or self driving cars.

 

[StatGuy2000]

Ah ok. That makes sense. So it would make more sense for me to do the Stats Phd. I guess I would have to take CS courses on the side as well. Because I think I would need to specialize in a topic in order to properly do artificial intelligence. Such as computer vision or self driving cars.

Please note that I was suggesting a possible pathway for you to pursue research in machine learning/AI based on your current background. I know for a fact that many statistics graduate programs offer flexibility to take CS courses on the side, and there is much overlap between statistics and CS in, say, machine learning.

That's not to say that you shouldn't try to pursue a CS PhD. That path is also open to you as well -- many CS graduate programs accept students from other related programs, and your undergraduate background in applied math is a good background for a CS PhD. You may have to take a few CS courses on the side, but I doubt you need to complete a second bachelors.

Try and talk to a few schools where you might be interested, or try and speak to advisors at your alma mater for some suggestions. Best of luck!

 

[FallenApple]

Please note that I was suggesting a possible pathway for you to pursue research in machine learning/AI based on your current background. I know for a fact that many statistics graduate programs offer flexibility to take CS courses on the side, and there is much overlap between statistics and CS in, say, machine learning.

That's not to say that you shouldn't try to pursue a CS PhD. That path is also open to you as well -- many CS graduate programs accept students from other related programs, and your undergraduate background in applied math is a good background for a CS PhD. You may have to take a few CS courses on the side, but I doubt you need to complete a second bachelors.

Try and talk to a few schools where you might be interested, or try and speak to advisors at your alma mater for some suggestions. Best of luck!

Thanks, I will look into it. I guess the issue is that most of my programming experience came from my stats courses so on paper it seems like I've been only doing stats not programming despite the extensive experience.

I've only taken 3 CS/Programming courses. One in C++, one in Matlab and one in Machine Learning. Although I have done bootcamp work in data structures and algorithms.

 

[Locrian]

I second everything Statguy has said here.

IMO don't back up too far, or spend too long in school. You're already on a good track to do this kind of work; grab the PhD, get some interesting projects/side-projects under your belt, and then get some real experience.