iujjkfjdkkdkf

From what I've seen, we have failed utterly in protecting vulnerable people. The focus on blanket lockdowns and enforcing silly rules has distracted from hygiene measures that actually could have kept vulnerable people safe. Where I am, almost all the deaths have been in nursing homes, which were a hotbed for covid cases. Fining kids for having a picnic in the park did nothing to stop that, but that is what people liked to focus on because it fits the overall culture that has formed around covid.

> In the US, they have – at best – protected the "non-essential" class from COVID, while exposing the essential working class to the disease. The lockdowns are trickle down epidemiology.

I wish this got more attention. The whole culture around covid is really centered on comfortable office workers talking about zoom fatigue and looking down on people that break the "rules" while the people that feed them are expected to keep working as usual, follow the rules about socializing and fun that have been imposed on them, and not complain.

If you haven't already, take a look at the detectron2 models and tutorials, especially the colab they have. It is a great way to get started.

https://github.com/facebookresearch/detectron2

I'm a big ML proponent, both because of work and personal interest. Saying it is like electricity is idiotic, and only makes false expectations about ML worse.

But I would not say that makes it uninteresting. Nor does it make it irrelevant from a business perspective. There are still lots of beneficial ways to use ML, and lots of different angles from which to derive intellectual satisfaction. I suppose if someone went into the field expecting I Robot type stuff, they might be disappointed, but it's hard to imagine someone with a CS background having such unrealistic expectations of what ML actually is.

Haha yes, saw it on my phone and had to hold it back from my face and rotate it back and forth to make sure it wasnt just my imagination.

I don't know if this qualifies as a "dark pattern", but this kind of "social proof" feature that the article is about buying really is hostile to consumers and is nothing to be proud about buying.

I don't see this solving anything. Paid work involves a meeting of the minds between the buyer and the person doing the work. This is why e.g. busking is not work, it is dressed up begging, regardless of whether the performers consider themselves to be working. This will just turn parts of the internet into more begging "like and follow", not produce quality content.

Yes I've found that too, and I regret I did not read more literature when I was younger. I'm objectively highly educated (phd in electrical engineering) but at some point realized I was pretty hollow outside of some narrow specialization - like Sherlock Holmes who doesnt know the moon goes around the earth because it doesnt relate to solving crimes.

To some extent I was not mature enough in high school to take literature seriously, but even now 20+ years later, I'm pretty sure I would reject it just as forcefully if it was presented as it was in school, with all the life sucked out.

Your point is much closer to the articles main question, which I understand to be "should you force yourself to read the classics" as opposed to "sould you consider reading classic literature".

Google's job is to get people to click on ads, and preserve their position by keeping a big portion of the world's elite tech talent well fed and docile.

I see this advice as mostly mapping to "get out side of your comfort zone", "do something everyday that scares you", "the definition of insanity is doing the same thing every day and expecting something to change", "you miss 100% of the shots you don't take" and so on. Overall it's good advice, only by trying new things and stirring the pot are we likely to get somewhere better.

This made me think a bit - I imagine most of the polarization these days comes from a vocal minority of either side, and that 95%+ of people are much more relaxed about politics (and still have a sense of humour) than we imagine from the media.

So it would be interesting to see if hyperpartisans could be separated from moderates by their appearance (just to be clear, I dont believe in phrenology, I imagine its information leaking from other aspects of presentation and background).

With respect to humour, the modern lack of humor seems to focus on not offending people, which I admit I dont understand but I could guess is predicated on the feeling that laughter is somehow equated to divisive mocking, as opposed to fun. I bring this up only because it reminds me a lot of Umberto Eco's "The Name of The Rose" where the 13th century religious leaders were arguing that laughter was inappropriate, and Jesus never laughed, somehow rooted in the belief that finding humor in things admitted the possibility of laughing at aspects of religion and therefore not taking them seriously.

Anyway, your comment made me think, so thanks.

I'm confused by the angle this "institute" seems to have about manipulating social / political discourse. (In that its weird to be providing these communication frameworks at the same time as actively using them for a agenda, but maybe I'm interpreting this incorrectly)

But that aside, the techniques that they discuss look like they could be useful for persuasive communication, including technical and product focused communication, so it is interesting from that perspective.

Thank you!

Has there been any study of variability in these activation images - like are there many disconnected local maxima depending on the initialization, or how they vary with retraining the network (or e.g. with dropout, etc), or varying the model parameters in some direction that keeps the loss in a local minimum.

I could picture that maybe they always look the same, but sometimes there would be cases where they have different modes that accomplish the same thing.

Can someone give more technical detail on what they are showing with the "neurons"?

They say "Each neuron is represented by a feature visualization with a human-chosen concept labels to help quickly provide a sense of each neuron", and these neurons are selected from the final layer. I don't think I understand this.

I dont think anybody's arguing whether companies should worry about theft. My earlier comment was that companies should not be using their concern about theft as justification for treating their employees like criminals, and subjecting them to bag searches.

Yes I fly, but I'm a normal looking white man, frequent flyer so I know how to pack and how to behave, and I've never had a problem with airport security.

I feel the same way about flying as in my earlier comment, nobody should have their dignity taken away to be able to travel, I'm lucky it has never happened to me in any serious way, and I support measures to end security theatre and treat everyone respectfully.

The money is important but I think it would be nice to find a way that people don't have to undergo this kind of indignity.

I'm normally a pretty libertarian person, but I definitely recognize how lucky I am that I dont have to be put in a position where I have to be treated like a criminal for the sake of money. Same reason I would never shop at Walmart, or take a drug test, etc. But I recognize that some people are not in a position to say no and get abused for it. And would be happy to see that change.

I would love to be able to pay for a professional GPS where I'm not the product.

Unfortunately, what Waze has going for it is the community aspect, I don't see a way that could be replaced in a paid product.

So Google may be ruining it further, but I can't see a way forward for a commercially viable "good" driving gps product, which is either a great opportunity for someone to build one, or just too bad.

Why do you think its appropriate for you to tell me what I can read?

I was thinking the other day about what books and movies I need to save for my kids before they are all censored. It's really sad that availability is limited by people's sensitivity, and now that so much is subscription based, it will only get worse.

I've stopped using amazon for anything unless either I already know exactly what it will be or if its throwaway and I dont care.

Even books now are often something that looks like it was printed on photocopy paper from a word processor.

I think amazon is converging to its niche (maybe not a niche if there is ~a trillion in annual revenue at stake) as a Walmart replacement. For instances where quality matters, I think they are already off the table for many people.

Not sure who you're responding to that advocated for that. But the closest I can see is the original post suggesting the that people join a protest group. Lobbying the government to force others to behave as you want, and not proposing an actual solution, is pretty close to demanding that a problem be solved without putting any effort in yourself.

I would argue the opposite, that its mainly a technical problem that has been coopted by a political movement.

The technical problem should be (roughly), how can we continue to enjoy our lives and the standard of living we have achieved, and continue to bring more of the world to that level, in a way that is sustainable long term. There are lots of promising routes to making this happen, and we have already seen lots of progress towards reducing environmental impact.

But there are people that want to move backwards, to reduce our standard of living, to reduce our freedom to make our own choices, and to enforce some kind of forced mediocrity that will help ration our natural resources. I reject this (and weaker versions of it) and believe there is more to be gained by focusing on progress than on telling people what to do.

Thanks, that's very interesting to hear.

I wonder if / how the concerns will change for different use cases, e.g KYC.

It may also be a question of educating users about the things they should care about.

Thanks for the answer. What I should have asked first is, do you think this is something that's important to your users, or are they happy to accept that there is research and engineering behind it that has concluded it is reliable and safe?

I'm asking because I've been in a similar situation (different area of computer vision) where nontechnical stakeholders are looking for assurances that the model is not going to fail under essentially unknown conditions. And same as you, we had some ideas, but it's hard to validate what best speaks to people, so was curious to hear if you had looked at it. Thanks!

Have you thought about what kind of assurances users / buyers are looking for about robustness to attacks or false positives, but also about working correctly under normal variability (a mentioned by others)? And how to best communicate those assurances?

Pour water into a maze (or create a potential difference across a conductor) and you'll see that as new current flows in, it doesnt explore every pathway but takes the correct path through.

This intelligent behavior arises from the simple compulsions things to reach equal potential.

The slime mold experiments are cool because they connect simple compulsions with emergent intelligent behavior in an organism. I have wondered if it's the same for us, if conciousness is really just the sum of all our simple compulsions, arising from basic rules - like is water "conscious" of wanting to seek it's own level, ions conscious of wanting to react, etc, and together that makes up what human conciousness is?

I think B is just a symptom of the complete failure of (many) business people to understand what ML is or what it is good at. It starts with people equating "AI" with something they've seen in a movie and working backwards, and like you say, ends with just a really bad way of interacting with a menu system. This is going to persist until either we invent AGI or AI stops being a buzzword used by business people who have no idea what the technology actually does to make up use cases.

A couple things: 1. as I understand, GPT-x are trained on generic datasets. Why would anyone expect it could just be repurposed for a domain specific task, without additional fine tuning? Same as a lot of vision models are trained on Imagenet and then fine tuned for the application, but you would never expect that imagenet weights would just automatically perform whatever specific task you had.

2. Where would this chatbot ever be a good idea? Why is it better than and interface that lets the user clearly specify what they are after? Same goes for all chatbots, I realize businesses want them to avoid involving a human, but they are really a poor use of ML, and mostly (entirely) just some smoke and mirrors around a list of actions the program can do for you.