Hacker News Reader

Comment by janalsncm | Hacker News Reader

I considered a PhD in machine learning. It’s mostly downsides. Granted, most fields are not like this but:

1) The field moves too fast to focus on a single thing for 4 years. A lot of people were devastated when ChatGPT essentially solved their NLP tasks.

2) Cutting edge NLP/vision research is being done in industrial labs as much as universities. And industry will probably outgun you with equipment (GPUs) and high quality data.

3) Pay sucks. You can make 3-5x working in industry. The opportunity cost could be a half million dollars.

4) You can get a lot out of a Masters in half the time or less.

View on HN

What sort of benefits come from getting a Masters? Everyone I talk to seems to say a Masters in CompSci is useless, and that you may as well do a PhD instead.

> What sort of benefits come from getting a Masters

Google will hire you to work on moving protobufs around as an L4, instead of an L3.

This cuts so much deeper than it has any right to.

By the time you get the degree, you could be promoted to L4 and saved money along the way, so you're saying it's a bad idea.

I believe a masters has helped me stand out as a candidate at least. Plus, I learned a great deal about computer engineering! The fundamentals have come in handy.

A lot of job requirements that I see ask for Masters or PhD, so you’re hitting the minimum requirement plus giving yourself a shot of having applicable work experience (read: doesn’t write spaghetti code). That said, there’s probably a huge selection bias due to my background.

A Masters can help with work visas and gaining residencies in some countries

Machine learning could be hardware as well. I'm in integrated circuit design and there are lots of custom hardware AI accelerators in development. Almost all of the new grads we hire have a masters degree in electrical or computer engineering (not computer science)

It helped me get a visa for Italy (I took part in a startup accelerator program).

Other than that not much else... I mean you learn things, but you could also learn things from watching YouTube, doing courses and reading books!

I've also seen people do PhD's and from the outside it seems like a lot of them did it just for the title - one of my friends though seems to love Academia and is now a professor.

As someone who hires, my opinion is the exact opposite. A masters is good, but a PhD is at least a yellow flag.

> PhD is at least a yellow flag.

Why?

Usually an inability to get shit done, too much of a scientist vs. engineer approach (not good in industry outside of very specific jobs), and a personality type that is not an aggressive go-getter.

Yellow flag not red because this is a general observation with plenty of exceptions, of course.

If I could go back to that age, I'd focus my PhD on actually understanding what's going on in ML models. Industry is always going to be incentivized to build things and not understand, so you can fill in the details. Plus it would be fascinating.

yup, if i went to do a PhD interpretability is the only interesting subject for academia IMO right now

It’s not like interpretability research is immune. You could’ve been in year 4 of your degree when Anthropic released their sparse autoencoder research. It’s just less busy because as you correctly note, industry mostly cares about getting the black box to print money.

> Plus it would be fascinating

You can do research on the weekend even if you’re not in a PhD. I’ve done it. And no one was breathing down my neck to publish it.

Reply on news.ycombinator.com