[cehrlich]

I gave up on a dream about a year ago. I had been working hard, not very successfully, for several years to make a passion into my career, and it not only caused me burnout and depression, but also made me hate the thing I was meant to be passionate about in first place. Now I'm on a mediocre, safe, boring path (web dev). Of course there is still some lingering regret, but overall I feel very relieved.

My advice would be simple: Try to take a few days off if at all possible so you can go into this with a clear mind. Then for each of your possible choices think about:

* What does a successful outcome look like here?

* What does failure look like here?

* Is there an in-between outcome?

* How do you feel about ending up in each of these cases?

* What is the likelihood of each of these cases?

* If you go all in on this and fail, what realistic backups do you have? Does this change your opinion on the previous questions?

Take however long you need to choose, then act decisively and look back as little as possible.

I will also say that your expectations of a PhD in medicine being way easier seems off to me, but I am clueless about both medicine and machine learning, so what do I know. Do make sure though that you are making your decision based on good information.

[rwanda]

Thank you for the thoughtful response. Switching to medicine is something I've been contemplating for a couple of months and though about all pros & cons, best-, worst-, average-, expected outcomes extensively, yet I still feel at a lock. To put it shortly, my best case is better staying at my current program and my worst case is worse. The average case is too difficult to determine as the probabilities I assign to each case are speculative.

As for a PhD in medicine being easier, I don't mean that doing the actual PhD is easier but rather the entrance requirements are. The medical school in my country is good so I can just do my PhD there which would be quite easy. On the other hand, the technical school I'm in doesn't have that much research in machine learning so I'd have to apply abroad which would make things much harder. Not to mention that machine learning is a hot area and perhaps the most competitive of all PhD-subjects.

[cehrlich]

Is your dream a PhD in machine learning, or to work on difficult machine learning problems? Many people (most, probably) come out of their PhD less happy than they were before, maybe there is a different way you can pursue this interest? If you even have a chance of becoming a PhD candidate, I assume you already have some skills in the subject that you could apply at a corporate job, on an OSS project, etc. I understand the appeal of "pure" research, but there are other paths as well.

I don't know much about the field so I can only guess, but maybe you can find some people here or elsewhere to talk about this who have the relevant experience?

[rwanda]

My dream is to improve & develop new machine learning algorithms. Weather I'd do this as a researcher or through a private company doesn't really matter to me. It just seems to be a prerequisite to have done a PhD to most such positions I see ( Take Google Brain as an example).

Unfortunately, there aren't too many people/research in my interest area in my country, but I'll try to ask my classmates. Thanks for your comment!

[marto1]

> working hard, not very successfully, for several years to make a passion into my career

But did you learn anything ? If yes, I'd think of it as a very expensive education.

[cehrlich]

In terms of quantifiable skills that your typical recruiter will be interested in, not much. In terms of everything else, absolutely. I also had much more fun in my 20s this way than I probably would have had otherwise, at the expense of entering my 30s broke.