While it would be insane to get a PhD if you had to pay for it, at least in the US you are paid a stipend (if not a good one) so you can generally avoid any debt. There is definitely an opportunity cost though.
I’m not so sure this is trivial and strongly depends on the biases of the agents that the group is composed of. It seems like some species of animals (humans, ants, wasps, etc) seem to select for intelligence in group settings more than individual intelligence. One of my favorite facts about human evolution is that in the process of becoming homo sapiens our brains actually shrunk relative to our ancestors possibly due to our reliance on social ties and specialization allowing for each individual unit to be “dumber” while the collective becomes more intelligent than any individual ever was.
It’s not apparent to me that current LLM’s have the right biases for a trivial unit of “more LLM” = smarter. It seems apparent that LLM’s are under many metrics more “intelligent” than the average person but they still don’t rival our corporations or collectives in terms of intelligence and I don’t see them being plug in replacements for humans and maintaining the benefits of the collective structures yet. Not impossible but definitely not trivial in my mind. We and our cultures evolved to work together and the modern world is the result of the emergent structure that resulted.
As a plug my favorite field relating to this is called stigmergy and basically describes how individual units like ants build intelligent collectives that are far greater than the sum of their parts.
Depends on who you’re talking about. How does the cost benefit analysis work out for someone while lives within a mile of a datacenter, owns no shares in the company, and pays rate fees for their utilities and will have to pay for maintenance for the increased load capacity as well as higher spot prices? If you view America as a cohesive unit I’d agree with you, but I don’t think these people are necessarily acting irrationally if you view their personal circumstance. It’s up to the building company to sweeten the deal and provide an incentive for the locals to add it. Based on the contracts I’ve read from many of these datacenters it seems more like they’re buying the pockets of corrupt politicians and developers in areas with high political inequality while conferring very little benefit to the people who deal with the externalized cost.
While I’m sure there are people profiting off these movements, there is very real grassroots support for blocking datacenters and generally anything relating to AI and I can’t blame anyone other than the companies building it. I think in 30 years this will be studied as the worst marketed technology ever. It’s one thing if outsiders try to say your technology is dangerous, it’s another if the companies themselves are telling the public that they’re technology is either going to a) exterminate humanity or b) eliminate all jobs and institute a permanent underclass. I think they’re right that these are concerns and while our niche tech bubble maybe thinking also about the potential scientific benefits of the technology, all the wider world sees is multi-national tech companies taking advantage of areas with high political inequality to push through genuinely unfair contracts that saddle utility rate payers with higher fees, potentially negative health effects, and with very little long term job creation. This is also coming from companies that have ignored complaints about the overriding addiction and division which their platforms have allowed to propagate over the last 20 years all while many Americans watch historic levels of corruption in government in favor of those same companies go unpunished. At this point I think the tech world has really burnt all of its political capital with the wider populace and while we may sit back and complain that progress is being blocked by luddites, its not irrational to want to block the creation of a technology which you will pay the cost for while seeing a disproportionately small percentage of the rewards. I think if we’re going to see the actual proliferation of this tech and the benefits it will bring, their needs to be a) more publicized news of real benefits to human quality of life in visible fields like medicine and b) an actual plan for how to confer the rewards of technology more widely rather than just continuing to center in the hands of so few people and not just approaching it as higher gdp but also lower wealth inequality as long as its correlated with political power
> while our niche tech bubble maybe thinking also about the potential scientific benefits of the technology
While all the academic talk about benevolent AI is fascinating, it has become quite clear that the development is driven by billionaires to become richer and big companies to get more revenue.
Everything else is just billionaires playing the tech bubble, so that they can rationalize what's happening and what they are part of.
Not sure what you’re talking about here. We can’t replace all energy needs with solar but it’s clearly one of the cheapest energy sources and with the added benefit of low capital expense to get started so you can set it up in distributed grids without the massive expenditure to support nuclear installations.
If it is profitable without subsidies from government then everyone would do it as there is money to be made. You do not need peer review studies to prove something is profitable.
Having dealt with the medical system recently, if there isn’t immediate harm involved with waiting, always advocate for yourself and consider a second opinion. The doctors will think you’re annoying and that you trust the internet too much but it’s their job to make you understand why elective procedures are necessary. Don’t let someone trying to rush to their next patient hurt your health outcome. I think in emergency situations this all changes though.
During my wife's second pregnancy, we got a lot more information and did more preparation than for the first.
Someone shared the BRAIN acronym for making decisions.
Can't remember it exactly, but it was like ask / think about benefits, risks, alternatives, intuition, doing Nothing (for now).
I think the Nothing one was the most important, what happens if I don't take action and defer the decision. Medical professionals often make you feel like a decision is urgent when sometimes it is not.
I like that rule. I’ve also found that for most things that are risky enough to warrant a second opinion it will probably take long enough to get in that you can schedule it several months in advance and then use the intervening time to learn more although calling medical offices to schedule appointments is no fun.
Plotting it this way is just a shift and rescaling. As the other commenter already showed this will lead to a plot that looks pretty much the same in terms of difference relative to the past data. I think standard deviation makes sense here since we don’t really have an innate sense of what is a meaningful change in ocean surface temperature and the point is to show that we are already far beyond normal here.
This is cool but I don’t know if the comparisons to conscious awareness really make sense here. Their definition of the J-Space is basically the expectation of how much a final logits output would change as a result of a small change in a particular layer (see past work on information geometry). This seems more to me like showing there exists an abstract reasoning subspace which is generally shared across different contexts. I guess you can relate it to humans but I’d prefer a more direct claim in a paper rather than having to present things in this more fluffy way.
Writing it honestly would defeat the whole point of it, that being, to push the narrative that their magical token predictor is conscious. They've been trying this for years now. This video is discussing a paper they published 2 years ago by the way... It's nothing new.
There’s debate over how much timing actually matters vs the rate of firing. Some people do believe in precise spike timing but I would say the general consensus is that spiking rate is a better measure of the current state of a system. There’s significant noise (as far as we can tell) in a neuron’s timing and it’s best modeled with a poisson process so we tend to think of it as rate coding which we can at least hand wave as viewing an RNN with a ReLU approximating.
Generally you can take a geometric view of this where certain features in a stimulus covary with neural activations in the same way they will with RNN “activations” which is at the real core of why people model things this way. The general idea being a dot product in an RNN can tell you something about what features are relevant for a task and we can look for hints of the same information being encoded in neural data. Certainly not everyone is in agreement on this though.
