Ask HN: Is PhD in AI worth it if I'd like to conduct my own research?

3 points by _xrdj 12 days ago

One acquaintance of mine, a teacher of informatics, offers me to enter PhD studies in AI at one local (EU) math/CS center where he has some connections. Because not many CS/math graduates currently want to proceed to a PhD and they're interested in AI and want more PhD students in this field.

He's the author of this [1] article on quantum fuzzy logic, which had received some international recognition, is obviously attached to it and would indirectly like me to build upon it. It seems decent to play with, but I wouldn't like to dedicate a whole dissertation to it, because there's more fundamental research to it. I'd love to apply to Michael Levin's lab at Tufts for some research assistance etc with some experimental ideas, but would like to build some minimal hardware and software prototypes first.

My background is in cognitive psychology/neurobiology and I have been teaching myself programming/CS during recent years. Initially I wanted to proceed to PhD in neurobiology and my supervisor recommended me for PhD studies. Currently I have developed several research lines in biophysics (and any CS etc required for it) for my own personal interest, and I was thinking it may be more efficient to start a business to fund my research and conduct it on my own or with relevant folks, rather than trying to fit systematized research attachments of others. Alternatively, I'd like to be paid for learning stuff I have a moderate to strong interest in, rather than hopping from one random job to another.

This teacher of informatics is prone to making generalizations about how the brain should work with such logic (though his logic was tested only on phoneme recognition) and doesn't seem to realize that <i>you are not your beliefs</i>. Can you imagine an AI generating such a statement, negating essentially any corpus of knowledge it was trained upon? And yet a human brain generated it.

Also from some indirect source I've heard that he's also supporting some racist war, though it has never been a topic of interest or direct discussion. And I wouldn't like to discuss my research interests full scale in such context.

Should I enter these PhD studies or better find a random job to continue on my interests in free time?

P.S. What do you think about this article on quantum fuzzy logic?


gardenfelder 12 days ago

Seems like only you can answer your own question. How you do so might be in the air,with unanswered questions like:

* Will that level of focus drag away from my personal interests?

* Why, in the long run, do I want/need the advanced degree?

* What do I want to accomplish in the really big picture,and what does the PhD project contribute to that?

  • _xrdj 12 days ago

    *Actually, if I enter in a PhD, it may only allow more time to focus on adjacent main fields of interest (electrical related biophysics/biotech). Conversley, if I get a random job, especially not remote one, it may distract more.

    *Currently it seems I do not need an advanced degree, because I know how to learn new necessary stuff, the basics of scientific method and can conduct the main stuff on my own. The limit is (formal) time and minimal equipment.

    *In the really big picture, hm, well, the answer may be ~1/137;). There may be multiple perspectives. Currently I focus to just obtain specific intended results in vivo/in vitro. PhD in AI may be adjacent to this, unless it would be directly related to electrical(main)/materials/3D/chemical engineering, but I only have general ideas about this rather than knowing which architectures/libraries to apply.

    • psyklic 12 days ago

      It sounds like you possibly aren't that interested in the field and have some doubts about the PI. What about suggesting a small research collaboration first, like a summer project or paid as a summer research assistant?

      It is very common for incoming PhD students to have research ideas that don't perfectly overlap with their advisor -- so the goal is to find some overlap that works for both. That said, if you come in hoping to research X but end up stuck on Y for five years, that likely will not result in a great experience. I would also be open about your other research interests -- in fact many advisors love PhD students who seek out collaborations. Otherwise, this might end up a point of contention.

      As for independent research without a PhD, that is certainly possible and many people do it as a hobby. However, a PhD would help a lot if you want to professionally be a researcher. Grant money will often require a PhD-holding PI, as will many research jobs. Some people might be successful without one, but I'd place my bet on those already publishing a lot pre-PhD.

      Regarding a business, people commonly start one type of business (e.g. consulting) with the real goal to fund something else -- a product/research/etc. This seems to rarely work out, since the business itself is typically more time-consuming than expected.

      • _xrdj 12 days ago

        Small research collaboration is a good point, I'll offer it for discussion.

        I actually don't know who would be my PhD advisor and what would be their research interests. I am generally interested in AI, but my main interest for a dissertation/experiments would be electrical biophysics/biotech, for which I've already collected some theoretical and experimental basis and which at least in the starting point may not require AI.

        Alternatively, I have ~20 ideas for AI, like "it would be cool to apply AI to speed up clinical trials" (there are folks at Foresight Institute, as I've watched their recent youtube presentations, who are doing that), but no knowledge of practical implementation details of these AI ideas. Or I would be interested specifically in AI used in Levin's electrical biotech lab or for epigenetic clocks etc, but there seems to be no such AI agendas in this center. in this center they seem to be focused mainly on recognition stuff, and I'd like a more generational/engineering aid AI. i wouldn't care would an AI work on quantum fuzzy logic or on a slime mold logic if it only achieves the intended result.

        I was not planning to do PhD specifically in AI, but it seems at least better than random jobs.

        I do not have any publications yet:). I have only 2 ideas for 2 articles I could write and publish on my website which I hadn't set up yet.

    • _xrdj 12 days ago

      The point is that there will be an interview in ~a month and I don't know should I disclose all my research interests, including AI related. Anyway, I would try to offer my own research agendas for my self at least, but as far as I know in the majority of these PhD doctorates one supports the other and may want students just to build in their already established lines of research.

bjourne 11 days ago

I suggest you delete this post. What you have written about this "teacher of informatics" is not nice.