Solid agree. Hallucination for me IS the LLM use case. What I am looking for are ideas that may or may not be true that I have not considered and then I go try to find out which I can use and why.
This technology (which I had a small part in inventing) was not based on intelligently navigating the information space, it’s fundamentally based on forecasting your own thoughts by weighting your pre-linguistic vectors and feeding them back to you. Attention layers in conjunction of roof later allowed that to be grouped in higher order and scan a wider beam space to reward higher complexity answers.
When trained on chatting (a reflection system on your own thoughts) it mostly just uses a false mental model to pretend to be a desperate intelligence.
Thus the term stochastic parrot (which for many us actually pretty useful)
Thanks for your input - great to hear from someone involved that this is the direction of travel.
I remain highly skeptical of this idea that it will replace anyone - the biggest danger I see is people falling for the illusion. That the thing is intrinsically smart when it’s not - it can be highly useful in the hands of disciplined people who know a particular area well and augment their productivity no doubt. Because the way we humans come up with ideas and so on is highly complex. Personally my ideas come out of nowhere and mostly are derived from intuition that can only be expressed in logical statements ex-post.
Is intuition really that different than LLM having little knowledge about something? It's just responding with the most likely sequence of tokens using the most adjacent information to the topic... just like your intuition.
With all due respect I’m not even going to give a proper response to this… intuition that yields great ideas is based on deep understanding. LLM’s exhibit no such thing.
These comparisons are becoming really annoying to read.
See, I never swore off beef, even with that. I took it as a reason to make sure they live their best lives, and to do my best to ensure that their deaths are as quick and painless as possible. I understand the hypocrisy or whatever that might be, but beef has kept my family out of bankruptcy when full time job income has not. I do apologize and thank every animal, for whatever that's worth. It doesn't feel great, but such is life.
Not to go full-PETA, here, but I think there's a pretty decent chance that in 300 years you and your family and this comment will be looked back on like how we now look back on remarks about chattel slavery.
Thanks for the compliment .. just so its clear, it wasn’t my leg, fortunately for me .. but it did definitely make me a bit of an indoor person at a young age. ;)
Money as "life energy" approach helps me be thrifty, but my real struggle is how do I asses risk when I want to buy an asset. If I wanted to use $500 from my work to buy a share of SpaceX, does it provide a better return than any other investment at some acceptable risk ? I don't see the life energy approach as helping this question but I need to read the book you recommend.
Note that SpaceX is not publically traded, but it represents assets I would like to own.
Its a paywall. Just I have to pay with info rather than money. It is not free.
To accomplish the goal of knowing how my congresscritter (really?) matches with me, I should not need an account. If I wanted you to contact them for me then OK but to call the site useless without an account just makes it so.
I don't work for PopVox, so I'm gonna stick with "Congresscritter" :)
Anyway, I guess you were expecting one of those hokey quizzes news sites have that never really cover the bases properly. shrug PV's actual purpose is to enhance constituent communications. That means your messages to Congress route through them so they can sprinkle on loads of metadata and make it easier for the Congressional aids to sort through while at the same time providing public accountability. This personalized bill-by-bill member-by-member thing is newly added on top of that larger main goal.
I use it because I like that accountability thing (transparency, woo!), and I have enough friends that used to have to sort through paper letters to the offices they worked in who have warned me that I will make staffers hate me if I send paper letters ;)
If all this is true, I would have a problem with a co-worker who would stand up and lie. By starting the meeting by being "super supportive" while knowing along with co-conspirators that you were trying to kill it, a good deal of creditability has been destroyed.
Maybe I am naive, and this is how things really get done, but regardless of which side I might have been on during this meeting, he is no longer trusted.
I kind of agree here. I think it says more about Googles culture not being all that different from any other big company. This article insinuates to me that politics are more important than sound ideas that are good for the business. If the skype deal was bad because it would need a large overhaul, why not just lay out a convincing case to that effect?
politics are more important than sound ideas
that are good for the business
The problem in big corporations is that there's lots of voices and lots of opinions floating around. You can't really distinguish signal from noise unless you have some kind of hierarchy.
Google cannot be different in that regard. They can't work with a flat organization in which Joe Sixpack has the same credibility as an early employee -- they are too big for that.
The only way such a corporation can improve (over the rest) is to build that hierarchy based on better metrics than the rest, to avoid situations in which big responsibilities fall on complete morons that can kiss ass. Such a corporation would also need to maintain the number of layers to a minimum.
Politics in big corporations are inevitable, which isn't necessarily bad, it's just a different environment than in a startup.
I think you're confusing politics with persuasion. Technical people have a problem with persuasion; I did, before I joined a large company, decided I wanted to actually change things for the better, and realized there's nothing dishonest about convincing people that someone will improve their company/lives. I was just bad at it.
Technical people seem to lay out arguments that are sound, but just aren't convincing or appropriate for the audience at hand. This is why shitty technology can win out when the customer lacks the time to make an informed decision, and makes the perfectly rational decision to go with the more convincing pitch.
And that's just it: decision makers don't have time to go through a thorough, completely objective, scientific argument. So, you either do what you need to get things done in what you feel is the right way, or you don't. There's nothing wrong with hacking a meeting, so long as the hack benefits the company, or at least is done in the company's best interest. As with software, the only bad hacks are those that benefit the hacker at the expense of the organization.
People rationalize why what's good for them and their allies is the common good. (A person who believes they don't, well, that's evidence against them more than it's evidence in their favor.)
I think there are some visionary things here but he misses the 'consent' part. By using free services we do give consent and if we don't like the privacy policies of Facebook et al then nobody forces us to use them. Yes providers should be honest about their policies and some do play dirty tricks. We learn which ones are not really honest so if we keep using them, we can’t really claim surprise or lack of consent.
Free stuff costs. Not money but there are costs. In the future world he is imagining there will be some who want to pay for their own machines and control the software that runs on them.
Being a parasite means riding a host and giving up some critical decisions about where and how the host lives.
By "consent", he meant "informed consent". That requires reading the terms of service and understand their implications. Few people actually do that. The rest of us merely click our way to something that "just works".
For instance few Gmail users are actually aware that Google is routinely doing semantic analysis of their e-mail. When I tell them, they're invariably mildly shocked. (I insist on "semantic" even if their methods are statistical, because Markov chains are highly correlated to meaning.)
The solution of including the cost of fire fighting in the cost of owning the house itself helps fix what I see as the mistake in thinking that I don't benefit from having my neighbors house fire put out.
If volunteer fire fighters are told by the people who they protect that it is ok to watch a house burn (because necessary costs were not covered somehow) then philosophically what is to stop them from not helping when I don't slip them extra money because their labor is volunteer anyway? I think this eats at the soul of the community.
Not necessarily, no. See for example p9, which documents two famous cases involving full-time professionals. Good screening practices seem to help a lot, but as you know perfect security is an unattainable goal.
My point was not meant to imply they would accept bribes. They do the job to fight fires so why agree with the idea that we should stop them when some equipment was not yet fully amortized? If they accept bribes or not, the idea you stop volunteers from doing what they want (fight fires) is wrong.
