This is a good suggestion, but it might be limited depending on what level the OP is interested in. I would expect that the OP would be able to be a part of research, but would be unlikely that they would be able to have much say in the direction.
Here are my thoughts, as someone who completed a physics PhD not too long ago and switched academic fields.
There are two parts to your question: 1) How do I study up on advanced topics efficiently and 2) How do I do research with my current circumstance.
For 1), I would first continue working through grad-level textbooks. If you need some structure look at the textbooks used for required course at your favorite/dream grad program/school. These will hopefully give you a solid foundation of the physics/math for current research.
The next part about specific papers is tricky, because a lot of physics (sub-)subdomains are highly specialized and have their own terms and notations. In general though, my process is to look at the references and citations, and sometimes the references and citations of those. I am basically either looking for a paper or two that frames it in a way that I have seen before, or points to an old textbook that I am not familiar with. The process takes a bit of time at first, but is smoother after doing it a few times.
For 2), it depends on how deep into research you want to go. The current path would lead well to being well-versed in the current landscape as an outside observer. I think in all likelihood to do anything more than that you will need to have some affiliation with a research institution. If you want to be able to pursue your own research interests within those institutions, you will very likely need a PhD. Unfortunately it is hard to give more specific advice without knowing more details/desires. If you are in the bay area, you could also look at LBNL or LLNL and see if there is anything interest there for you.
I personally don't think so. I see roughly two categories:
1) Professional type degrees such as health related occupations (medicine, dentists, nurses, etc) and engineering (mechanical, chemical, etc). These are occupations where you more or less need a relevant degree, and the salary has more or less priced in the cost of that degree.
2) Jobs that require a degree as a screening mechanism. These are jobs that do not require highly specialized skills, but pay better than local average/better than minimum wage. There are usually a lot of interest in these jobs. A college degree acts as a filter to reduce the hiring pool to a manageable size. A lot of people in this group see the cost of college as a type of investment (large, upfront investment, long term higher annual salary).
Coding related jobs is somewhat an exception to these groups. While a lot of big companies are more likely to want a degree, there has historically been a lot of opportunity for a self-taught developer to find a job. I think the vast majority of jobs still fall into one of the two buckets, however.
> I like this, as scientists need hypotheses to test and often those can come from non-scientists.
Bold claim, or tortured use of often.
> I wonder how many of our scientific discoveries have come from suggestions (or bold statements) from people in other fields.
A lot of academic fields are fairly siloed from others. One example is presentations at a conference from a subfield that is just slightly oblique from yours is hard to follow because they use different jargon than you. It is getting a little better with some emphasis on interdisciplinary research.
> Sure, many may prove to be false, but I like how you've framed it as a novel opinion that could spur further research into a topic.
From my experience with being blasted by spam ideas just because my contact information is in a university's physics directory, a lot of/most 'novel opinion' is noise.
Looking into blast waves might be of interest. The jargon of the difference between a shock wave and a blast wave somewhat depends on the sub field of interest.
Agreed! I am currently planning out building a similarly spec'd machine (or waiting for zen 3), and trying to pick out a motherboard that is a good balance of features has been a headache.
From my experience, it is a pretty simplistic take.
A lot of people enjoy the social aspect of college (from parties to having a cohort that is interested in the same things they are).
The money issue roughly fell into three buckets: 1) students not worrying about money because they have rich parents, 2) students not worrying about money because they have student loans, and 3) students worrying a lot about money because they have student loans.
From my experience going to undergrad at a state university in the midwest, a lot of students would prefer to have a well paying job immediately after high school and not go to college. However, a lot of those same students would rather go to college than stay in their small hometown where job prospects are bleak.
> A lot of the problems in academia, I believe, come from incompetence, a lack of questioning or at least of doubting others' competence and your own competence, and reliance on unsound ideas about scientific methodology.
I kind of agree, but I will state it somewhat differently. Note my experience is in physics and healthcare, so may not apply for all fields.
In my experience, the desired skill set shifts to more management/admin/bureaucracy/money-chasing once your in a professor or professor-like position, as opposed to nitty-gritty researcher in the grad school phase. The incentives for the grad school phase is good science. The incentives for the professor-like phase is grants/papers/awards.
I am not remotely in this field, and have not been following this closely at all. With that being said, what obligation do they have in maintaining GPT-2? Did they have some stated commitment that they walked back, or am I missing something else?