therobot24

the bitter lesson is to _let_ data and compute do the majority of the work, it's not to remove humans completely, but wherever possible

Have you tried google deep research? i'm curious how well perplexity compares to it.

I've been using Gemini Deep Research to replace my Google search since it does a web search and provides links you can check yourself to any citation that the resulting report uses.

until there's actual enforcement, there isn't the incentive to tell the truth...

It really is sad how much data has been captured and monetized of the average person. It seems like we're only continuing to turn up the heat as we continue to 'boil the frog'.

republicans wanna spend the way 'they want to' and cut every way the democrats want to spend, and just the opposite is true too (austerity only really matters when it's the other party with the control of the budget)

pair this with each district, state, and constitute company/nonprofit/whatever looking for a piece of the pie creates an utter mess of non-stop spending

I think you're focusing too much on the analogies he used instead of the thesis -- everything is aligning to a leader who isn't respected outside of the USA, much like kim jong un. That if saber rattling is the future of the US, then divestment will only increase and won't easily return. And further, the benefactors of this transformation is whoever doesn't have to adhere to our saber rattling which will be as many other countries as possible.

would rarely put "high quality" and "tesla" next to each other

https://www.jdpower.com/business/press-releases/2024-us-init...

I think the writer should have also taken a moment to reflect on the underlying _need_ that led to these revolutions. Each need is implicit to what's discussed in the table, but would be useful to help frame the discussion of the underlying questions that are discussed. For instance, going from IR3 -> IR4 created a _need_ of information management, where Google was the first major player in providing search as that means, now agents are taking the role of search & retrieval as a single action.

ok i think i'm starting to get it a bit, the double edged sword is essentially the reason for focusing on emergent results which (i'm guessing in Friedman's mind) have more options for mechanisms for good results than a constrained market might.

> In my view, the scenario and reasoning by Friedman in how he applies market forces applying to business decisions is a view where you have an assumption of 'knowing the result'

I think it only appears that way if you take small quotations like this out of context.

Can you help me understand this. Like the purpose of using logic to deduce decision making is because there's a fundamental structure in assuring the result from those deductions is accurate (or enough to continue manipulation of the problem data). When you try to predict market reaction to a decision and ascribe harm or success based on the decision you're creating causation out of correlation, often incorrectly. Again, not an expert in economics, but when i view theories of free market forces and how they're part of the logic of business decisions i can't help but think the people using this information are assuming their knowledge and logic aren't fundamentally flawed and as a result are essentially just guessing without thinking they are.

So my interpretation of how Friedman drew his theories of market (and granted I'm no expert, just a layman reading things online) are about every business should optimize as much as utterly possible even if this results in exploitation, because the market will accept exploitation until it can't and correct. I think you and i are aligned in this belief. Where i think we're separated, likely cause i just didn't explain it very well, was that to me, this means you as a CEO are accepting exploitation under the guise that you have the ability to react to the market forces to correct for it. That you 'know the market' and can adapt as needed. Such as described in the Friedman quote within the article which expresses an extremely libertarian view of how companies should be able to hire and fire workers:

“...consider a situation in which there are grocery stores serving a neighborhood inhabited by people who have a strong aversion to being waited on by Negro clerks. Suppose one of the grocery stores has a vacancy for a clerk and the first applicant qualified in other respects happens to be a Negro. Let us suppose that as a result of the law the store is required to hire him. The effect of this action will be to reduce the business done by this store and to impose losses on the owner. If the preference of the community is strong enough, it may even cause the store to close. When the owner of the store hires white clerks in preference to Negroes in the absence of the law, he may not be expressing any preference or prejudice, or taste of his own. He may simply be transmitting the tastes of the community. He is, as it were, producing the services for the consumers that the consumers are willing to pay for. Nonetheless, he is harmed, and indeed may be the only one harmed appreciably, by a law which prohibits him from engaging in this activity, that is, prohibits him from pandering to the tastes of the community for having a white rather than a Negro clerk. The consumers, whose preferences the law is intended to curb, will be affected substantially only to the extent that the number of stores is limited and hence they must pay higher prices because one store has gone out of business.”

In my view, the scenario and reasoning by Friedman in how he applies market forces applying to business decisions is a view where you have an assumption of 'knowing the result' of any decision and using that 'knowledge' as justification for reasoning. When in reality, you don't know the result, you can estimate, but that's about it. So when an exec is applying Friedman principles they're trying to 'know the market' and that's a fundamental error in my mind due to the chaos of the world how it can manifest across all avenues of life.

optimizing business value in the manner perpetuated by Friedman assumes you know more than you do and can adapt to things because of that knowledge, ultimately no one can actually do this because the world is chaotic and difficult to anything and everything that tries to master that chaos

That's why you use gov to fund reliable research, collective money funding collective good of knowledge

paywall

If only it were that simple, Marc Andreessen's interview in the NY times put it best. The wealthy in tech were hurt by the anti corporate youth movement that resulted after the 2008 crash. They (the wealthy) didn't understand why so many employees were so against them, they then saw the Biden gov continuing that anti corporate stance as DEI and other initiatives to potentially control AI and crypto. The tech billionaires hitched their wagon to Trump as a way to 'survive' regulation and saw another opportunity to compete with each other through who had most sway over the most easily bought president since the teapot dome scandal.

Gov job in tech is like any other job, the best employees who can work together successfully to solve problems is best optimized over a wide area (spatially). There are other benefits of strong employee rights and stability in work, ability to transfer to other departments a pretty simple process (which opens the type of work you can do while still being employed so much easier), and good health/retirement packages. These are motivators for people to want to stay with the Gov when already hired in, but to new and existing talent that have more options, the lack of remote and/or telework can be make or break. If we actually want a Gov that can perform, because people do that work and thus would want good people, we should be not artificially constraining ourselves on how to achieve that goal.

payment isn't bad, just need to make sure everyone gets their cut

i only performed a quick read of the paper but couldn't find how many humans they used to generate their expected human performance, this seems to be the main content:

To ensure that we did not overfit PaperQA2 to achieve high performance on LitQA2, we generated a new set of 101 LitQA2 questions after making most of the engineering changes to PaperQA2. The accuracy of PaperQA2 on the original set of 147 questions did not differ significantly from its accuracy on the latter set of 101 questions, indicating that our optimizations in the first stage generalized well to new and unseen LitQA2 questions (Table 2).

To compare PaperQA2 performance to human performance on the same task, human annotators who either possessed a PhD in biology or a related science, or who were enrolled in a PhD program (see Section 8.2.1), were each provided a subset of LitQA2 questions and a performance-related financial incentive of $3-12 per question to answer as many questions correctly as possible within approximately one week, using any online tools and paper access provided by their institutions. Under these conditions, human annotators achieved 73.8% ± 9.6% (mean ± SD, n = 9) precision on LitQA2 and 67.7% ± 11.9% (mean ± SD, n = 9) accuracy (Figure 2A, green line). PaperQA2 thus achieved superhuman precision on this task (t(8.6) = 3.49, p = 0.0036) and did not differ significantly from humans in accuracy (t(8.5) = −0.42, p = 0.66).

look at those for loops! should look into fft-based correlation, can even do so with melon transform for scale and circular harmonic transform for rotation

IVAS is still too clunky and not worth the ROI to get that level of funding, expect it to get canceled unless congress greatly increases the Defense budget or changes acquisition practices

I'm a clinician who works in child development pathology, which is why I commented.

sounds like you take research at it's word instead of understanding the fundamental ideas and concepts that are being explored...classic white coat thinks the book is right and everyone else is wrong

well....yea you can be high performance and _just_ focus on what you're interested in, but agree - pushing yourself to be leadership without being a supervisor to _build_ 'the high performing culture that can deliver something great' is really special

you're not wrong that the result will likely be somewhere in the middle, but i think china is an excellent case study of how much you can really influence and change behavior economics at scale

it's because he could afford it

As these modeling systems are becoming increasingly complex, it is hard—and getting harder—for a graduate student or postdoc to 'come up to speed' quickly enough to really understand the full scope of the model development needs and wrap up a development project on the typical three-year timeline of a proposal," said David Lawrence, who co-leads the Community Terrestrial Systems Model at the National Center for Atmospheric Research. "Unfortunately, that leaves many projects unfinished."

Click bait ad for funding, move along

to whom? zoomers (https://techreport.com/statistics/discord-statistics/)?

Anything with 100% accuracy is suspect, either via the model, the dataset, or the means of measuring

AI (LLMs etc) exhibits "emergent behavior" properties. things like reasoning and other higher order effects.

I would disagree with this statement. The model is _reasoning_ via the method of processing the data since the weights are informed from the relationships of prior similar information, but at its heart the model is still just doing pattern matching. The primary distinction and what I disagree with in your statement is that these AI models are closer to AGI because there aren't a bunch of if/else statements or basic pattern matching.

i think their choice of interface has definitely put them behind compared to their competition

whenever biometrics pops up on HN i always have to post the reminder that a biometric is _both_ a username & password bundled as one login credential. People like to compare biometrics to passwords, but that's a bad analogy because passwords can be changed whereas no one in tech likes to admit that a username should be changeable too.

not crazy with some of the results, e.g., https://www.bulletpapers.ai/paper/1d002187-927d-6775-94e2-a4...

- Bulletpapers title: Using robots to map and digitize construction sites

- Paper title: Multi-agent robotic systems and exploration algorithms: Applications for data collection in construction sites

- Bullets / Key Details: + Proposes methodology for multi-robot systems in construction sites + Robots use exploration algorithms to navigate autonomously + Information from building plans guides exploration + Robots digitize environments by 3D scanning as they explore + System is robust, efficient, requires minimal human involvement

- Generated Summary: This paper proposes using multiple robots with different capabilities working together to map and digitize construction sites. The robots use exploration algorithms to autonomously navigate and scan the environment. Information from building plans helps guide the exploration. The multi-robot system is robust, efficient, and requires minimal human involvement.

This all reads like the info was gathered from the abstract instead of the paper itself....that said, this is good AI generation info for IEEE explore to implement i guess