dontreact

My own personal take:

I live my life not for enjoyment but to try and leave the world better than I found it. If I can succeed in doing so (even slightly) then teaching 2 children to do the same would be my life’s greatest contribution.

I actually think the declining birth rate is rooted in the narrative that humans are ruining the earth. I believe we are making it better and more interesting over time.

I spent some time listening to some Spatial Audio mixes about 1 year ago. Dolby Atmos is a cool concept but I was not able to hear a benefit.

This reminds me of how when stereo first came out artists (including the Beatles) didn’t take it seriously. They would spend most of their time getting the mono mix right and the the stereo mix was thrown together and would often do strange things like having the lead vocals on one side.

Does anyone know of artists that have really taken the time to ensure their music is benefitting from Dolby Atmos?

Aren't there units smaller than nm?

I would not make the same inference because we know that ML systems generally struggle with robustness and the long tail, while humans tend to be exhibit much more flexibility to adapt to distribution shifts or unusual situations. For a concrete example of this in computer vision, see work such as the Imagenet—C dataset where simple distribution shifts generally tank ML models but do not impact human performance.

But regardless, the claim here isn’t “under some assumptions that some people (and not others) find reasonable, we can extrapolate and predict that self driving cars will be found to be safer”

It’s “self driving cars are safer”, which there isn’t enough evidence to claim yet.

To me safety is mostly about the risk of injury and death to me. And mostly about death.

Humans die from driving only once every 100 million miles on average.

So until there is a comparison at this scale, to me it’s a very incomplete picture of safety to say that you are outperforming humans.

If driving my own car means colliding 100x as often but I expect to die 2x less often, I would consider self driving to be much much much more unsafe. There is really no way for me to understand the risk of fatalities from a sample of only 7 million miles, since either humans or a system with 2x the fatality rate of humans would both be expected to have 0 fatalities at this scale.

Given we are at 7m miles, hopefully this comparison is coming soon and I will be much more convinced.

I know many many people who have moved to SC from the Bay Area. I mention the proximity because it’s relevant with regards to people becoming familiar and then desiring to move there (whether retired as you mentioned or working remotely), and not because it’s a commuter town.

You have to remember that Santa Cruz is very near the Bay Area and in California. A lot more than 60,000 people want to live there many have been forced out by costs.

What you’re saying really amounts to saying you want to stop immigration into your town. While I can understand the sentiment, the problem is that people feel this way in so many places that it is creating a crisis of housing affordability.

If you want to really dive into the math and motivation for this entirely new class of models (state space models) I highly recommend reading Albert Gu’s thesis.

https://searchworks.stanford.edu/view/14784021

I tried to read the Mamba paper and I was lost but after reading his thesis it was a lot to understand it.

The part I’m still struggling with is the section on how Mamba trains so efficiently, it’s surprising since it is actually autoregressive and step by step and I thought this is why LSTMs were dropped

If installing these things required passing FDA trials before you could even do the first install, it’s hard to imagine anyone self funding the regulatory trial and then open sourcing the result.

These businesses you mention are different because you can start getting paid as soon as you build some expertise in how to operationalize the systems.

Yeah… but people sometimes dont do this

Im not saying it’s always nitpicking. I’m saying that sometimes it is.

I generally like the tool.

When your code review tool is really nice one unexpected negative is that it can create a culture of nitpicking. Sometimes it’s not worth arguing over small details like variable naming but the tool makes it really easy for things to head in that direction.

Sometimes variable naming isn’t a small detail. But sometimes it is and it’s a waste of everyone’s time to argue about it.

I used to feel the same way until I called a Waymo and it picked me up and dropped me off in SF, with no driver.

I looked through a bit of the companies here and I don’t see any companies that have to retain a quality system and staff to stand behind their products permanently. These models seem to work better when you can just put stuff out there and occasionally pop in to help?

This is not a model that can work with regulated medical products. There is a very significant cost to maintaining static artifacts and I don’t see how you can defensively do that if anyone can access the artifacts?

Open source is a good model for science but how will this lead to regulated medical products?

Yes there is an unregulated layer but I think it’s clear the FDA will certainly care about the most impactful applications of AI in healthcare

To deploy generative AI in healthcare someone has to pay for the salaries of a lot of people to do the work. That means there needs to be a business model.

I am not sure who will take an AI through regulatory procedures if it is open source and there is no way to make money from it.

Open source is a useful tool for research yes. More of it would be nice.

But I don’t understand how or why anyone is going to go through all the hurdles of deploying technology if all of it is open source.

Maybe an open source enthusiast can explain to me how that is supposed to work?

Your understanding is not correct. You can either a) schedule a pickup using any app or b) stroll in and place an order off the menu. A) is way more convenient so I haven't done B) in more than a year though.

I live next door to one of these (mentioned in the article) and I think it's an improvement to my neighborhood. This article seems like a huge over-exaggeration of some minor issues to me.

It's nice to be able to have a ton of food options in your neighborhood where you can go and pick up as well.

Not if you catch it early stage. With LDCT screening you can catch the cancer when it is very small and physically remove it.

I did and it was wrong. I was responding to the claim that they got a plausible result.

Depends on the market. I have heard from someone who I trust deeply and knows the financials well (worked there) that Uber is profitable in certain markets.

It’s very possible that they have to pull out of a lot of places, but I think the model can work in the right places

def calculate_relative_advantage(interest_rate, loan_size, market_return): # Your calculation logic here pass

Chat gpt actually implements it

I tried this question and GPT4 did way way better to getting closer to a final answer. Phind was horribly wrong. I can't help but think something seems off with your eval given just how badly Phind did on this.

I want to make an interactive plot in Colab where I can show

X axis is interest rate of a 15 year mortgage. Y axis is the relative advantage of buying a house vs. renting in terms of total net worth at 15 years.

Assume a monthly budget for renting + investing or buying a house of 10k

Plot different lines for a few different market returns.

Make a slider that controls the total size of the loan.

This is nice because convolutional models seem better for some vision tasks like segmentation which are less obvious how to do with ViTs. Convolution seems like something you fundamentally want to do in order to model translation invariance in vision.

It is in fact for many people a solid argument for a wealth tax.

Yes it's difficult. I think we should have a pretty high bar for evidence required to decide that this idea is impossible.

Because once you decide that it's impossible, then there will be no more evidence collected.

I would like to see a wealth tax passed and then evaded and then a follow up to stop the gaps etc. A few cycles of this before deciding it can never be done. I think this is a multi decade effort that is important for the future of all societies.

Ok but I can't change the fact that everyone is greedy. What I can control is whether I support nuclear, which seems like a better tradeoff for the future than not supporting nuclear.

We are all human and have flaws and it's best to work around them rather than idealize about a world that doesn't exist.

But the waste is just SO small and just needs to stay inside of concrete to be considered contained.

I would rather burden future generations with this rather than continuing to produce CO2 at the scale we are doing now.

https://en.wikipedia.org/wiki/Onkalo_spent_nuclear_fuel_repo...

It feels more solvable than trying to figure out how to reduce CO2 without nuclear (to me)

Well... they hardly tried all genres :)

It sounds like it can't handle lyrics or semantics that well so I suspect any genre where the lyricism is important would also be quite mushy and recognizably AI