It boggles my mind that an LLM "understands" and acts accordingly to these given instructions. I'm using this everyday and 1-shot working code is now a normal expectation but man, still very very hard to believe what LLMs achieved.
But even if they can theoretically be hacked, so far Waymos are still safer and more reliable than human drivers. The biggest danger someone has riding in one is someone destroying it for vindictive reasons.
A naive question: So much “tax payer” money is going towards research funding. But it looks like private companies are reaping rewards, mostly as a new drug. Why is this research not (mainly) privately funded?
It's impossible to discover a basic fact, such as "the mitochondria is the powerhouse of the cell," and monetize it fully within the same organization. A million scientists can take a look at that basic fact and involve it in their own research in ten million ways.
It's also not practical to keep those facts as trade secrets over the several decades over which their applications need to develop. Even if an industry consortium was willing to discover that clouds are made of water droplets, it would certainly leak before the science of meteorology had progressed far enough for that consortium to offer saleable rain forecasts.
Finally, companies are unwilling to train people about basic facts. Academia is the only system where "and then you tell everybody" is a part of the incentive structure. Privately, you have a strong incentive to reveal nothing and punish leakers.
that's a pretty funny example you gave, because the discovery of the chemiosmotic effect was not funded by the government, it was privately funded by a guy who raised money and holed himself up in a regency estate with an assistant for a few years to prove it.
Private and public philanthropy both contribute to science in the same way. If you're asking why private philanthropists can't replace federal funding, it's the same as the reason why we can't just make billionaires pay all the tax: the NSF budget would bankrupt Bill Gates in 13 years.
More realistically, what would happen would be that rather than sacrificing themselves for the greater good in some kind of voluntarily socialist outpouring of wealth, they'd ask us to look to China for our scientific future.
Stop doing expensive science? Some things can just wait for tech to catch up and make science easier by lowering costs.
An example is the superconducting supercollider. Chemistry and pharma industry were making NMRs cheaper and the cost of the supercollider components went down. So the LHC was a much more effective "buy" than the SSC.
Basic research creates foundation knowledge that can drive medical innovation, but rarely does academic research create final composition of matter. Private funded work is all the non-research components of drug discovery - optimization of molecules, regulatory work, commercialization, etc...
To imply that private companies reap the rewards of basic research without contribute much is ignoring the many other components of translational work.
To give a more financial answer, it’s because pharma products have a low probability of success and have long lag times. That means a high cost of capital: lenders and investors are going to expect good returns to make up for the risk.
Second, biotech/pharma actually already do invest quite a lot in R&D. But they tend to focus on translational work rather than speculative exploration, because it is less risky.
a lot of basic research is very risky and most of the time it’s not stuff that leads to immediate development of a new drug. it’s basically acquiring knowledge with the hope that some of it might turn out to be practically useful in the future, but in the short term, it just allows us to understand stuff. but it’s not directly profitable, so private companies aren’t motivated to invest so much money in that
this sounds like it's a really bad idea for the government to fund. What's to stop someone who happens to have made it in the ivory tower go crazy, spin up some kooky ideas that are highly risky and just blow taxpayer money on something not really accountable?
Most research is funded through grants. Many different federal agencies provide grants, as well as private organizations. When applying for a grant, you have to indicate what you're going to spend the money on. And your grant may be rejected if the organization funding the grant thinks you're not going to spend the money well. And if you can't find someone to give you a grant, you probably can't do the research, even if you have tenure.
There are problems with this system. Researchers often have to spend a lot of time writing grant applications, and grants can be rejected for any number of bad reasons. And there are cases where research was funded that probably shouldn't have been funded. But research funding isn't given willy-nilly to whoever asks, and taxpayers wasting money on kooky ideas isn't a particularly big problem.
I understand your reasoning, but our efforts to prevent this is part of why professors spend so much time writing grants and filling out other paperwork these days. It’s better in my opinion to just accept reasonable risk.
I would add that weird ideas can be surprisingly useful; nobody expected research on gila monsters to lead to our most successful weight loss treatment to date.
the fucking gila monster story is pure revisionist history. back in the late aughts after the human genome got sequenced and qpcr started picking up it was pretty obvious from islet alpha cell proteomics that glucagon would be an interesting drug target.
researchers don’t receive unlimited funding for life, even if they made it into a permanent position. they have to regularly apply for grants, and those applications are reviewed by experts and have to be grounded properly in previous work. it’s just that potential for profit is not a criterion for evaluation, as it is in the private sector
That's why when you apply for grants, they are reviewed by the panels of experts and then you have to report the results/progress. Nobody will just give you money for something crazy.
Who should judge scientific quality, if not individuals familiar with the methods used to generate the results? We are living the counterfactual right now, where political opportunists with axes to grind have replaced the pattern matchers you describe.
nobody is saying this is not the least worst way to fund science in general. the point is that use of taxpayer money demands a higher level of accountability that this method cannot satisfy.
And yet the grant writing process continues, as the perfect shouldn't be the enemy of the good. I'm glad someone at the DoD thought that ARPANET was a good idea to research so that 60 years later we can argue online about whether the govt is just giving out tax payer money to whoever for any reason.
what a tired old argument. you dont know what would have happened if DARPA did not fund ARPANET. we might have had something better. we don't live with access to reasonable counterfactuals.
Au contraire, this is a tired armchair reasoning argument.
We don't know what could've happened, but we do know what did happen. It's like those people that say that the New Deal was bad, actually, and if we did nothing that would be the same or better!
Right... but no. Because the New Deal did pull us out of the depression. It's one of the most potent and effective pieces of policy in American History. We can play armchair economist all day. But we have to face what we know worked and think about why it worked.
In part because the expiry of patents puts a cap on the return on investment private research can get you.
Patents last for about 25 years, but important innovations have returns far into the future, hundreds of years. At that rate, you would very often be better off accumulating interest on capital anyways.
Notwithstanding the nature of scientific progress as an accumulation of smaller experiences (each individually harder to justify with a profit motive).
Indeed, even privately funded research is often openly published, such as the now-famous paper "attention is all you need". There's just not that much to gain from keeping every single thing under wraps. More to gain with openness.
Even ignoring the limits of patents, how much of this research "pays off"? Do 1% of research grants go towards something tangibly useful, or is it closer to 0.1%?
The timeline from preclinical work to a new drug application has a ~5% success rate. A major bottleneck in this is target selection, which research should in principle help with. Giving a number for how much science improves this is iffy, because a lot of research is in fact pointless, with tiny specks of gold. Overall, when comparing money spent vs. its effects (ROI, economic spillover, cost-savings, etc), its definitely worth it.
A growing suspicion of mine is that maybe the govt is just more efficient at some things, like funding speculative research, which is part of the industrial policy of every prosperous nation.
While there is value from having the drug becoming available on the market, you would think there would need to be some form price controls in exchange for private production.
Depends on your role. Assuming you are their direct supervisor, simply give more tasks. Slacking off is usually due to not enough to do. Follow the tasks at a high level, make sure he knows he’s being tracked. This has even the effect of increasing motivation, as most people value attention.
If slack off means he’s not showing up at team meetings, constantly missing deadlines, low quality of work, etc.. just give feedback on a case-by-case basis, ask how he would make it better next time.
If he fails to improve and the examples pile up, time to make hard decisions. I have to say, in my experience as a direct supervisor, only 1 out of like 4 people fail to improve, if you are willing to invest your time and effort.
My experience is the opposite, when people is unmotivated or not interested in the work, its unlikely for them to improve even if you give them enough chances.
I have the impression that people get in some kind of comfort zone when underperforming that's psychologically hard for them to improve until they move to another team/company.
So, you should catch these issues as soon as possible to let people know that this is not tolerated, in some way, this helps because people can take you more seriously.
I think it could be just bad fit. Work looks good from outside but once someone is hired, they get disillusioned because it turns out they don't actually enjoy business domain or whatever else and find the job unfulfilling.
I prefer spending some actual quality of time (having a walk, having a conversation) with my SO instead of staying idly next to my SO and half consider it couples time.
Right but if we're dealing with time constraints then it's makes sense that watching something both of you are interested in _together_ is a more efficient use of collective time than watching it separately and talking about it during another activity.
If you share an interest then go for it. What I meant is that some folks prefer to prioritize watching something "meh" on TV with the excuse of spending time with your SO and then complain they don't have time for gaming.
Instead of half assing two things: free time and familiar time. Spend some free time on your personal interests and then spend time with you SO where you are both engaged with each other!
This is actually bad advice for the "marketing professional" case for two reasons in my opinion;
* Author directs his client (?) to create hypothesis out of thin air and start testing them. This is how you loose time and money. The missing step in between is observation before hypothesis creation.
* In a workplace setting, the #1 thing you should do is to ask for feedback from your supervisor, incase you feel something is wrong or you have a performance problem and don't know what's wrong.
