Tell me about PhD's, please.

In My Humble Opinion
1

Specifically in Computer Science / Artificial Intelligence.

I wasn’t sure whether I wanted to carry on studying a while ago due to my course being pretty dull, but now I’m in my penultimate year, the course has gotten a lot better and pretty much convinced me I want to go on to get a PhD. What little research I have done (a research essay on robotics, a mini literature review for a handwriting recognition project) I have enjoyed.

However, I’m confused about certain things. I’ve been looking at university websites and reading about their admissions policies. Most list a set of research topics that are currently being studied in the department and then state that any of those can be studied or you can suggest your own. What exactly do they mean? Is it just a case of me finding a supervisor willing to take me on for any subject, even if it’s not currently under investigation in that department?

Secondly, after reading these admissions websites, one recurring theme that keeps popping up is “prior research”, i.e. I must be able to show research that I have already carried out and get people to vouch for how good it is. Now, my course (a BSc) isn’t heavy on research at all apart from a few essays (like I already mentioned). I was thinking that I could join the British Computer Society and write for their “Computer” magazine on some topics. Would this be acceptable, or is the research they want more like published papers in a journal (of course, it depends on the school, but I’m asking about the average case).

Thirdly, I’m confused as to how PhD’s are funded. Do you get some sort of wage? Do you need to take out loans, etc?

Thanks for the help.

If anybody else has anything to add, please do so, regardless of whether it’s related to CS or not.

2

I wasn’t in CS, so I couldn’t answer your other questions, but grad students (at least the ones I knew) generally got a stipend. It came out of a grant or fund earmarked for purposes that could include student support. You generally have to do research or teaching to hold up your end of the bargain. Nobody gets a free ride. I don’t know of anyone who had to pay their own way through grad schoool (as in med school), but it probably happens, too. Talk to the folks at one of your prospective grad schools.

3

Longtime CS prof here, but in the US not the UK so Y[del]M[/del]KMV.

Generally there are 3 kinds of support available (not all places use the same terms, etc.).

  1. Teaching Assistantship. This can mean anything from actually teaching classes, to running labs or recitations, to grading papers.

  2. Research Assistantship. Working under a prof. Lots of work, little to no credit.

  3. Fellowship. Usually not required to do anything, but also doesn’t pay much in most cases. Some schools require people with fellowships to still Do Something for various reasons. In the US, NSF fellowships are the most famous, but there’s a lot of others available.

In CS, there can also be other jobs available, such as working in a computer lab, doing programming work on a project, etc. Usually not tied to other goodies like a break on tuition.

A common pattern is to work as a TA for a couple years then as an RA for the remainder, but that’s not always the case.

In the US, at the top schools, it is quite rare for someone in CS to have to pay their way thru. If you are good enough, you will get money.

Note that it sounds like you already have adequate “proof of research.” They really don’t expect undergraduate theses. Make sure a prof you worked with stresses this in his/her letter of recommendation.

One thing to consider is how devoted are you to a particular area of AI. If you are absolutely sure that you love one area, then apply to schools that have profs working in that area and let them know why you are applying there.

But a huge percentage of CS grad school applicants put down AI as their area of interest, become disenchanted, and end up doing something else. If you aren’t sure of your devotion to the topic, then the overall quality of the program is important. Are there other things that you would take up if AI doesn’t work out for you?

I would forget trying to write articles for that magazine, it probably won’t help and could hurt.

The Big Point. Your current profs know a whole lot more about the situation in the UK than any of us. Talk to them about all this.

4

If you’re really interested in Artificial Intelligence, you’d probably want to go more into the computational linguistics end of things. Read up a little on informatics, ontology, and expert systems.

And yes, my understanding is that the people who want to go to grad school are supposed to get in good with a professor and work in that professor’s lab, while they’re undergraduates. That’s the way it worked for me.

5

I should have made clear, I’m already studying AI. My degree is a joint CS and AI degree, so I have at least some ideas about what I’m interested in and what I’m not.

Thanks for the advice. I’ll probably speak to my Director of Studies about this sometime soon, see what he has to say.

6

Why do you want to go on to grad school? Is it because you kinda, sorta like research and you can’t think of anything else better to do? Is it because you’re dreading a 9 to 5 job and liked being in school? Is it because you liked the college-ey lifestyle and want to keep living that way for a few more years?

If your motivation for going to grad school is something other than either: 1. getting specific credentials for something specific that you really want to do, or 2. a true, total, can’t-live-without-it love for your subject, then do something else.

I can’t emphasize this enough. Grad school isn’t just a job or a school program–it’s a lifestyle. One that can really, really suck and leave you pretty poor. Realize that the costs of grad school aren’t just the tuition and fees your TA or RA waiver covers; you’ve also got some pretty heavy opportunity costs to consider. You’re giving up at least a few–sometimes as many as 8 or so–years of at least a semi-decent paycheck and greater opportunities for leisure. That opportunity cost can start to feel pretty massive, especially as the years go by.

I strongly recommend that you have some source of major emotional support be before you start grad school. Grad school can be a very emotionally taxing and isolating experience. You’ll need some pretty intensive moral support at times. I’ve noticed in my program that grad students with committed, long-term SOs or close family members in town tend to be more successful in their programs more quickly than single people do. Having a household with another person offers some important economic help, besides. If you’re a full-time grad student, you probably won’t be earning much; having someone else sharing your living expenses and pulling in a second income doesn’t exactly hurt you.

I’d also strongly recommend that you try and talk with the other students in your prospective department before you apply. Or, at least, before you accept an offer. How happy you are or aren’t in your program will depend an awful lot on how much you like the other people in your program. People who have good experiences in grad school are people who form friendships with the other grad students. If lots of people seem really unhappy, or if you can’t see hanging out with the other grad students outside of class or seminar, go elsewhere.

Make sure that you and your advisor will have a good relationship. Find someone who is or was in the lab you want to work in. Take that person out to lunch or for a beer or something, and ask them for a candid description of what it is or was like to be in that lab. I wish I had done this myself, before I signed on with my present project.

7

I apologize for the awkwardness, typos, and crappy grammar in my last post. I’m exhausted, since I’ve been teaching like crazy recently (I’ll have some sort of entertaining Pit post sometime relatively soon about the crazy lecturer I’m working for this quarter) and desperately need a good night’s sleep.

8

As someone who went into a Ph.D. program in astronomy, and dropped out with a master’s due to issues with my advisor (among other things), I can only say

ABSOLUTELY TRUE

to everything in Scribble’s first post.

9

My experience couldn’t be more different from what scribble said. I have a PhD in computer science (25 years old, so beware) and did it at two different schools. I moved because my advisor died on me :frowning: :frowning: :frowning: and no one remaining was working on what I wanted to work on. I had passed all my quals and stuff already.

I have lots of contacts with professors, have spoken at schools and have been an attendee at a graduate seminar at a local, very famous university, so I don’t think I’m too far out of date. My daughter is about to apply to grad school also, but not in CS.

First - you have to enjoy research (which it seems you do.) You can get through without loving it, but it won’t be as much fun. I do, and I had a wonderful time in grad school.

Where I went all prospective PhD candidates got funded. RA ships were usually higher prestige than TA ships, and let you do your dissertation work while being funded. The crucial thing is to join a group. Most RAships in the US are funded by grants, so you have to more or less pretend you are doing something relevant to the grant. I got an RAship immediate in my first school because I was interested in a topic my professor was working on, and few other people were. But I have dumb luck. I also came from MIT, which helped. Many others started as a TA then got an RAship when they found a topic and an advisor. You need to show some ability, though.

Now for topics - there are broad areas and specific research areas. In the broad area, you really need to go to a school where someone is already working in the area. If you like AI, and go to a school without an AI program, you are going to get lukewarm support at best and be pressured to work on a subject being funded most likely. Within an area, it depends on the professor. I had no trouble getting to work on the thing I wanted to do, which was within the area being worked on by my professor. I think many students never come up with an idea and get assigned something.

As for ttg’s comment on being an RA - not my experience. Maybe it was a lot of work, but it was work I wanted to do and loved doing. I’ve had the advantage of doing mostly research type stuff in industry (rare these days) but nothing was as much fun as being an RA. You don’t have to worry about the details of the grant, you just get your money and explore. Credit varies by professor. My advisor never was first author on papers from his students, and I never, never had any complaint about getting credit. I know some professors who act like their students don’t exist, but most really care about giving their good students visibilty.

I spent the first half of my graduate career single and the second half married. I don’t know that one was easier than the other. My wife did type my dissertation into Multics for me during the day while I coded (I wrote it in the night) but I felt more pressure to get done than I would have if I were still single.

I do agree that you should talk to students and to professors who you might want to work with.

You might look for other things besides AI to work on. That is so popular, and so last millenium. But that is a different topic.

Oh, and some of the best advice on how to succeed in grad school I ever saw was given by Fred Hoyle at the beginning of his novel Ossian’s Ride.

10

Not sure why Sattua is especially promoting computational linguistics in particular. However, you probably do want to select a field where you can eventually get a job. And since the late 80’s, early 90’s, AI has taken pretty much a backseat to other areas of computer science. For better or worse, universities tend to hire researchers in whatever the trendy research of the moment happens to be (in the early 90’s, it was networking) And, because a PhD can take 4-8 years, you can’t predict what’s going to be hot when you’re graduating.

So my advice would be to understand why your undertaking a doctorate and be realistic about selecting a topic that will get you hired when you finish. Selecting a “hot” topic will also help you find a professor with decent funding.

11

I am currently a PhD Student working in AI.

The field of CS and even AI is huge. No university is large enough to cover even a remotely comprehensive sample of it. So what they do is that each unviersity forms labs or groups who specialised in a certain particular area and concentrate on that. Different universities will have different specialisations. While I suppose it would be theoretically possible to propose a topic outside of a particular uni’s specialisation, I would highly suggest against it. Research is done in an enviroment and that enviroment matters.

Research is research, not journalism. I’m not sure how the British system is structured but in Australia, most students are required to do an “honours” year in the final year of thier undergrad in which they conduct research. If that’s not normal in the British system, then it probably won’t scuttle you in the application process. However, if that is the norm, then your going to be seriously
disadvantaged if you don’t do it.

Having said that, if your planning on applying to a program, speak to your prospective supervisor first. They usually have quite a bit of weight in whether you get in or not. If you can impress them, that might count for a lot.

AI is becoming a huge and booming sector in the last few years. The term “AI Spring” is being bandied around since the late 80’s were termed an “AI Winter”. Furthermore, AI is a lot more cross-disciplinary than previously. Many other fields are requiring the use of AI techniques and knowledge of AI can lead you to a bunch of different directions in your research. Do you have any specific interests in mind? I’m not initimately familiar with the British scene but I could probably point you to some well thought of unis if your field is close enough to mine (Computer Vision and Human Computer Interaction mainly).

12

an anecdote:

I once talked to a Phd student from my department…unshaved, almost staggering. He barely opened his glazed-over eyes, and muttered, “well, I had to decide…either I was going to enjoy life, or I was going to get a Phd.”

13

Paging Angua.

Actually, since she mainly haunts NADS, you’d be better off asking there.

14

I didn’t know that this was also going to be a thread about how horrible grad school is…

Actually, it was the best time of my life. Sure, lots of work, low pay, etc. But it was all fun, exciting, new, etc.

Other comments:

Don’t ever pick a field just because based on job availability. Do what you’re great at (which should also be what you like doing). There is always a need for people who are great in any given field. The worst thing you could do in college is major in a field you don’t like and aren’t very good it.

As to current uses of AI in business. It goes in cycles. A few years ago, “data mining” was the big thing. Now- biometrics. By the time you graduate maybe it’s the bottom for AI again. Winning at this kind of game requires being psychic. So don’t sweat it.

15

I feel the exact same way. I got a PhD in Physics and I loved grad school. It was a lot of work at times, but I was able to make time for an active social life. Also, because grades mean very little when you’re getting a PhD, you can focus a lot more on learning on the material and less on cramming for exams, unlike in undergrad. And once you’re done with classes and just working on thesis research, life is sweet. Unless you hate what you’re doing, in which case you’d better switch advisors/fields asap, because it won’t get any better.

16

I was planning on doing my PhD at Manchester, if I can get in, which is a lot closer to home and within driving distance of my family. Manchester’s also got a strong reputation and has quite a big AI research program.

Yeah, I’ll be doing the honours year next year, complete with my project. What I was worried about was a) all admissions pages seem to want two references from people familiar with my research, whereas my honours project will be under the tutelage of a single advisor b) the project that I’m doing this year for my AI course is being marked anonymously by another person in the class, as my instructor is from Japan and only speaks pigeon English (don’t ask :rolleyes:), so I was wondering how I could get the second reference.

I was thinking about going into robotics.

No, it’s because I love computer science.

Seems to be lot’s of conflicting advice in this thread, some people are advising me to go for what I like, others are advising me to go for where the jobs are :smiley:

17

And is a fantastic city. :slight_smile:

Chances are, you’ll be either funded by either the Engineering and Physical Science Council, or some private, industry related grant. I’m fairly sure that Manchester probably has a lot of links to industry, and will have a couple of companies sponsoring postgrad research. The way that’ll work is that your tuition fees will be paid, and you’ll get what’s called a maintainence grant, which, if you’re research council employed will be about £12k/year – enough to live on certainly.

The best thing to do is to look at Manchester’s website (and the websites of any other places you’re interested in), and email the members of staff whose research most appeals to you; get to know what the projects would be about and stuff like that.

If I’m reading things correctly, you’ve still got another year to do yet. This is good. Contact Manchester now, as they may have some summer projects on offer, which they may consider you for (and pay you for). If you manage to get onto something like that, you’ll be at a great advantage – you’ll know the staff, you’ll know the department, you’ll know if the field’s for you, and most importantly, staff won’t be relying on your references (they really count in this sort of thing, so make sure you get on well with your project supervisor, and that you work hard in your final year project – working with established, recognised names helps a lot), because they’ll know you, and whether they can work with you, and that makes all the difference.

Hope that’s helpful.

18

A PhD means very different things across different fields and you should try and get a handle on what a PhD means in CS and AI, particularly what it means from an employer’s perspective (I guess you’re doing that with this thread). In some areas a PhD is simply the last degree you sit on the road to being qualified, my own area of chemistry is a text book example of this. Anyone who wants to do have intellectual input into serious chemistry needs a PhD, in a nutshell. I get the feeling that this is far from the case in other areas, though. A PhD can be seen as a very scholarly, isolationalist pursuit in some disciplines that can hamper you in the job market. Anecdotally, a friend of mine has a PhD in actuarial science and got hired despite his PhD, not because of it; it was seen as a weird thing for a guy to have in that line of work.

A good way of totally bolloxing up your job prospects is to be not particularly good and get a PhD. I think its important to attempt an honest appraisal of one’s talents before starting a PhD. Very difficult to do because its not about what you’re good at now, rather its what you can become after three years in graduate school.

Scribble makes good points above, but be aware that many UK PhD programs are three years long. The catastrophic eight years of purgatory that can happen in US grad schools cannot take place in the UK for a science PhD.

19

Really? Wow!! I don’t know anyone–not even the brightest among us–who managed a PhD in just 3 years. Are you talking about all the grad work needed for a Phd, from just after undergrad through graduation, or the number of years you need to put in after getting a master’s?

If it really only takes 3 years in the UK, I’m all for relocating. It took me 3 years of field work (and, yes, I needed to sample in all 3 growing seasons) just to get the data for my master’s thesis. How do I sign up for grad school in the UK? And could I, as a US citizen, get funding from a UK institution?

20

The way it works here is that you do a Masters generally a part of your undergrad. When you do postgrad work, its pure research. There are, generally, no taught courses, no quals or anything like that. You go in on day one, maybe get a few weeks to orient yourself with the subject, and then are thrown into pure research. Most people, apart from the really intelligent ones, rarely finish their PhDs in 3 years – 3 years is normally how much funding you get. Most people take a few months over 3 years. I only really know of 2 people who finished their PhDs within 3 years.

As a non-UK/EU citizen, getting funding might be difficult, unless you get a private studentship through a company or something; the main grant giving bodies tend to reserve their money for UK students.