What would the pain-points be, in your view? Is it that the LLM is generally not “intelligent” enough, or is there some other limitations? As someone who enjoys using LLMs for self-education up to the bachelor level, I’m curious where doctorates might face issues.
From my experience using LLMs and supervising PhD students, the LLM is too focused on solving a task rather than educating or opening paths. It’s great if you have a specific question to ask, even at quite high level, it can explain you group theory or quantum chemistry quite well if you ask the right questions. The problem is that if students knew the right questions they would not be students. LLMs can get stuck explaining a topic without realising that it would be useful or interesting to branch out to something else. They can give you all the pros and cons of a given simulation technique without even trying to tell which of those actually matter in a specific case. They tell you if you ask, but you have to know what to ask.
LLMs are not great at open-ended discussions and don’t benefit of experience in the same way humans do. They are better to solve problems than to explore ideas. They keep getting better, though.
Your PhD advisor would be lending their wisdom and intuition to get you pointed down a promising (novel) research path. I don't think LLMs have been demonstrated to be at human level for that.
How are you using for self-ed? Would you please link to something you have found useful? Any time I've tried to do something basic like use an included learn skill/style or roll my own variant, it doesn't click, and I end up just doing what I would have done 10 years ago: watching lectures, reading the book, etc.
I tried to do a project on Claude.ai to self-learn cs 15-213, gave it a skill, pointed at the cmu site, put the labs and my notes up on GitHub, but ended up just reading the book, doing problems, and asking questions as needed. I struggle to engage in long-form work in this manner, because the bot is always pestering me about the course itself. I could probably just write "don't ask me about coursework" and that would fix it. Poor Claude always wraps an answer to my one-off question with "would you like me to add this to the week x materials?"
I find the cost of my writing the question, waiting for the reply, parsing the llm response and translating it to something that makes sense, understanding it, and cross-referencing, is too high. I suspect I'm holding it wrong.
wasn’t there, a year or so ago, some sorta OpenAI plan where you’d pay $400 or something and get “PhD-level research“ capabilities? I wonder what happens to that :)
go look at the forums for ubiquity unifi pro routers. they have the worst behaviour caused by capacitor residual power. they will not restart properly until they are discharged. so if you have a powercut they will not reboot.
you have to unplug for 30 mins, short the power cable pins and then it might reboot
the good news you can see the fake pricing on the amazon website by just looking at pricing history - they literally put the button at the top. Every sale is a scam.
i dont understand the concern. they are putting up great financials. you have to invest ahead of the outcome. this is just classic quarterly public company earnings BS, where public markets dont reward innovation investment. they just want crank the handle financials.
The bigger issue is on the model front, can Google compete; Gemini doesnt seem to be able to compete on the heavy expert end; they are doing well on lighter faster models.
I've given up on Gemini. It sounds smart but most of what it tells me ends up being wrong or misleading. I might actually hand $20/mo to OpenAI. It's been far more helpful with the random collection of legal and health problems I've thrown at it. My recent comment history is going to make me come across like a shill for them but, holy crap, GPT has been doing amazing things for me at work as well.
I don't get it either... Google has so much talent yet they just can't seem to get it right.
At work it's paid but in my own personal queries it's free vs free and gemini seems to give erroneous answers with confidence frequently and they will completely flip flop in subsequent queries of the same text. It seems very unstable.
People bought Google for the torrential free cashflow, that looks like its gone forever with this new capital intensive model. If that the case then it needs to be valued like a heavy industrial rather than a capital light tech company.
> this is just classic quarterly public company earnings BS, where public markets dont reward innovation investment
Genuine question; but aren't these treating stocks as speculative and on vibes? One can say that these comments could be true for the first signs of cracking of dot com bubble. Sure, Web eventually succeeded but many tech giants from dot com era (AOL/Yahoo and so many more) eventually went to dust for spending too much time on the innovative bandwagon.
During the Dot-com bubble really tried to give this example but IIRC there were companies like pets.com who lost 2$ for every 1$ of sale so how a company treats its financials do matter a lot.
The market doesn't seem to reward innovation sometimes because there have been times the first persons to innovative have actually really failed to capitalize on that innovation and many extremely innovative businesses like Airlines (We can literally fly speak of innovation!) have been terrible businesses investment-wise generally speaking.
I agree Gemini's value is not at the frontier, but they are making very useful smaller models. 3.5 flash lite is super fast, cheap and token efficient and accepts any kind of input. For large, price-sensitive processing, in consolidated workflows that don't need the latest improvement, it is the best. Perhaps they should focus on improving their dominance on this business, I don't see gemini reaching claude/gpt in coding. But they could be the ones that make it possible, and economically viable, to roll out llms in large scale automation.
seems to me that we are at the asymtote for most usage. sure run the prompts that need the frontier on generic silicon but burning the fable 5 model into silicon could be perfectly viable
reply