Part of the problem is that so many places have no plan for integrating new hires at all. They drop them "near" their team and expect "magic" to happen.

Actual planning and ramping up new hires I believe would really help with this remotely - you'd have scheduled meetings, you'd have a teams / slack / whatever chat running, you will have a project board or ticket system or whatever.

I don't think having people who need help just wander up and interrupt someone else who's working - potentially on a complicated issue - ever actually worked well for the companies. It killed productivity.

You know, T-Mobile and AT&T have some of the easiest ability to use resellers if you have paid your phone up and get it unlocked. You just swap the SIM. Number transfers over and everything. And you can actually set the pre-paid to very low cost if you have low usage, or usually still less cost if you are a big data user.

Oh, and most of these sell PIN cards at a bunch of gas stations all over, as well as at the independent cellphone stores out there. So... if you have to pay in person, you're probably even more convenient to grab at your local gas station. If you can pay online, you can buy them on ebay at a known / fixed price without extra fees.

So, I guess this helps with google search traffic, but certainly not engagement.

What, exactly, has reddit done recently that shows they understand large masses of their users or the Internet in general?

Ehh, the EVF is higher resolution, and a lot easier to handhold stabilize than with the LCD and holding it out in front of you - also easier to see outside in bright sunlight. The only reason to use the LCD is if it's fully articulated and you need to do a different angle shot, or you've got it stabilized on a tripod.

IDK, I guess people I know weren't at all worked up about the deaths - because in context that was 2 wars (I thought Afganistan was sub 5k Service Members), contractors no one I know had much sympathy for (i.e. they not only signed up specifically for *that war* but got paid big bucks compared to Service Members who usually signed up to *defend the US*, which these wars weren't). But that may just be my circle.

But over ~ 20 years? We lose 5x as many people to car accidents *yearly*. It was to my circle basically 0 causalities. In my perception we lose close to as many service members yearly in training accidents and testing Ospreys etc. I know it's not exact, but no one I know was "We're losing too many troops". I guess when your uncles were in Vietnam, *that* was a war where we lost people and we really wanted to pull out because of those losses.

I met a lot more people who were like - we could have universal healthcare for that wasted money. We could have much better educational subsidies. Hell, we could fix the roads.

I'm not an expert, but I did hear on the Embrace the Void podcast # 157 I think about GPT3 (and GPT4 is just a bigger one, no major change to architecture) that they do have an understanding of the semantics of words via semantic vectors and may be doing better than humans at that level of understanding.

I'm not sure how you could answer questions and react to feedback without some "understanding" of what you're doing, but maybe you mean it in a different sense?

For instance, I understand how to drive my car and operate the pedals and such to get myself from place to place, but I don't deeply understand what the transmission is doing.

Even in language, I can often use a word given hearing it a lot in context, but not necessarily be able to give you a deep history of the word or give a good dictionary definition.

Presumably people have to get involved eventually or else you just have a bunch of robots blowing each other up in a field or something. I would think for the war to be won, you'd still have all the issues that humans on the defending side would still fight back. And then, just like with the US wars in Iraq and Afganistan, it wasn't really killing the soldiers there that caused us to pull out, it was all the equipment costs, fuel etc going on "forever". We lost "hardly" any troops.

I mean, it kind of seems obvious that we want ads / shelf prices to be what you pay at the counter or at checkout, but we want our "cart view" and receipt / invoice to detail the line items so we can know if we're looking at $20 shipping on that $5 item and decide we want to look for a better deal on shipping or just go pick it up in person etc...

And some of this is more wanting smarter / better versions of like expedia where you want to specify that you want to fly on day X and will want 2 checked bags, to pick the seat, and travel insurance for instance, and then see the list of options priced FOR THOSE THINGS. It's pretty unhelpful to see ticket prices when there's some that include a checked bag, some that don't include picking your seat beforehand, etc.

I'm sorry but that is a statement of faith not science.

I guess you could phrase it that way. Do you think the brain is not doing information processing via physical processes?

This is essentially the same universal function we plug into evolution on the notion that eventually it can't accidentally arrive at any possible answer and it has had billions of years to do so. We aren't building these systems to operate over multi-billion year lifespans.

I don't really understand what you're saying here, or the relevance to the rest of the discussion. Maybe clarify?

We also no longer have a scientific understanding that the portions of the brain which inspired these systems give rise to consciousness.

I haven't mentioned consciousness, mostly because we can't define it or really explain what, if anything it's doing, or if it's in fact post hoc rationalization in an emergent phenomenon. Clearly you don't have to be human level consciousness to produce art - look at all the animal, and now machine generated art.

I just think that at a fundamental level, humans are doing the same sort of thing computer systems are, just that it's more complex. In terms of "same thing" I mean information processing, not some very specific set of steps. Do you think there's a important difference between a traditional chess program and alphachess? They both play chess, often better than humans can, but one is programmed by humans and one is AI self taught. But at the basic level, they are inherently computer programs running to play chess.

I just think it's the same sort of thing when humans are doing information processing.

I think there's more interest in society supporting education than landlords, but that's just my perspective. And this is a big difference between a 17 year old and a large developer corporation. Also, landlords have a physical asset they can obviously modify or sell if they screw up.

I don't think they're actually comparable that much.

And yet the training photos have been identified from AI output because the AI incorporated bit by bit identical chunks

I'd love to see the information around that. How big is this chunk? Because there's a threshold where obviously it's not an "identical chunk" by any reasonable analysis, yet technically it is an identical chunk. What I mean is, if it's one pixel - cannot be a copy / taken chunk by definition really. Or it loses meaning and plausibly could be "taken" from about everything. But in high res images, IDK, 128x128 pixel chunks might well fit that. Consider just the existence of a gradient etc. But now it gets fuzzier -> Look at image compression algorithms. That changes the bit by bit copy, probably quite a bit. Is that now "not a copy"? If you reject that "sidestep" then it seems like it goes the other way too. Having a set of eyes isn't a copy or lack of creativity. But there's a finite (if large) amount of photos with eyes in them. There's a lot of similarity between two given set of eyes in an image, that's what makes them a set of eyes. How different does it have to be before it's considered not copying?

This AI isn't outputting plaster casts, those are exact copies of a whole sculpture. From someone on the outside, this looks very similar to giving a brief to an artist and getting 4 or 6 options to pick from or refine further. The artist doesn't have unlimited creativity, they have to produce something that fits the brief. But the artist surely isn't just plaster casting there either.

Yes, the human painter is influenced by everything they've seen before but the vast majority of that isn't even artwork.

Ok, and so your argument is that a bigger training set is needed.

The painter has self-awareness, tastes, preferences independent of ANY input

All of that is basically random transforms adjusted by external inputs (life experience). That is more or less relevant for spec creations, which is what people are using these AIs for. It just seems to be a difference in complexity. This just seems like an odd "god of the gaps" argument in reverse - because we can fully explain what the AI is doing makes it less impressive. I just think that's wrong - I bet eventually we'll fully explain what the human brain is doing in a similar way to the AI explanations, and that won't fundamentally change the creativity of the brain just because we can explain it.

I just think there's no reason to think painters aren't inherently doing the same thing the AI is - they're influenced by any art they've seen in the past. They are taking very small bits of style and what they think a window would look like in this situation, and mixing it together and making a somewhat original output.

Lets take fantasy content, something there is no "original" for the painter to see. Their representation of a dragon is just a remix of all the other drawn or painted dragons they've seen.

When it produces a different image on the next request the basis for that variation will be the result of systematic adjustment of internal weights from past input, not a creative element.
You think human painters don't make systematic adjustments from past input and feedback? They certainly take reactions to past work into account!

I might be more materialistic than you, but I'll say the main difference between AI and a painter in the creativity area is just that humans are still more complex math we don't understand as well yet. At it's base, it's still just physics happening.

I'd say just like it doesn't matter whether you send a file via FTP or bittorrent even though bittorrent breaks it up into tiny little bits in random order from many source locations, and FTP just sends it in chunks in a linear order from one source - you still get a copy of the file at the end. And here, you still get a unique new image at the end, whether a painter does the mixing and different seeds and randomness with a brush or an AI does it digitally to an image file.

Heck, we can't even define art when humans do it - think of inkers, animation artists who did the in between frames, and more - the banana on a wall, or the urinal on a stand. Art is such a generic term that anything can be art IMHO.