coldcode

It only works if the background is continuous and fairly static - you can't easily synthesize a complex background if you have no reference material. You can of course extrapolate the visible background using other AI that has knowledge of real surfaces; however if the background has familiar content (like a sign) that is not visible in the reference frame(s), it's unlikely to fool anyone no matter how well the face is animated.

I was in Silicon Valley just before the start of Dot-Com with nearly a decades experience running a couple of software startups (in Texas, not in a high tech city) working at Apple and left to go back home thus missing out on enormous opportunities. I thought Apple would go out of business and figured no point in sticking around. What a doofus... I never looked around and asked, would my experience be useful somewhere?

My career was OK but who knows where I might have wound up.

I worked in a financial company in mid 2000s where the network head did not believe in internal firewalls so that all internal users were on the same network as all the web app servers and database servers. If someone was downloading a movie then customer web access slowed; since everyone used Windows everyone was required to run virus scanners on their computers and that included the app and database server machines since they were not isolated from the rest of the network. If a vendor came to demo something they were unable to since there was no way to isolate their laptop from everyone else so they could not access the internet.

Good thing I never put any of my money in the company accounts...

I find SD to be amazing technology, but it still (mostly) sucks at producing "intelligent" images. It basically fancy math that turns noise into images (from the opposite it trained on) but still has no idea what it is producing. If you run it long enough you eventually get lucky and find a gem. I like to try "George Washington riding a Unicorn in Times Square"; I've so far never gotten anything a first year art student can draw. I wonder how long it will take before something more "AI" than "ML" will have an understanding even close to what a simple human brain can process.

In the meantime it's fun to play with it, plus I'd like to better understand the noise training process.

The thing about having crazy monetary ideas—eventually money will go where it wants, and you are left looking like an idiot. You can make all the claims you want, but an economy is witnessed by everyone, and you can't fool them all.

Is it really all that hard to build a TV designed to be used only with a separate device (Apple TV, Roku, etc)? I wonder why no one has started a company to do this; perhaps VCs only want to fund people who make money with ads.

I think this is probably more common than we might think. Given remote work being preferred lately, this is probably doable, especially if companies are hiring large numbers of random contractors. I know my old team members (I retired recently) have gone through a ton of remote contractors recently, many of whom were completely useless. Thankfully I only hired people before Covid hit, and at that point everyone was required to work in office (at least for the work I was responsible for). Everyone I did take on was either excellent or at least competant. I did hire most of them off of phone interviews only, so maybe I was lucky.

If an experienced person does the interviewing asking the right questions / requiring tests / etc might be insufficient to realize the person you are interviewing is not the person who will do the work. I wonder how you would catch this before actually having the "worker" start.

I guess this is a downside of all remote work assuming your company is less than thorough in checking references/documents/etc.

Welcome to the world of software development. Only at my last job (~ 6 years before retiring) were there huge projects that were inevitably late. These were projects for the company's business, not external. Typically we were handed some requirements with a hard deadline, then everything changed daily (and was required for day 1) until we neared the deadline, then the deadline was moved; rinse, repeat. Then after months of this sometimes it was cancelled, sometimes it was postponed further, sometimes it finally shipped. It was a dance I got used to dealing with. Longer hours were asked, sometimes mandated, but everyone knew it was part of the show, and while as lead I worked longer hours I refused to demand my team did. In the end it didn't matter, as these projects involved huge numbers of people/teams so blame (fair and unfair) was spread around.

Once something shipped, if it worked, then no one remembered the pain until the next one.

If everyone is working long hours including execs and it's shared pain, something will eventually change for the better. My problem with it was when I worked long hours but the execs went on long vacations and weren't even available for approvals or questions. Then nothing will change and you should go elsewhere.

I only look at art tweets and tweet my new work several times a week. Other than that I ignore all the crap.

When I was at Apple in the mid 90's (Before Steve) I remember the Mac OS source code had tons of special app checks, mostly for various Microsoft (also others) apps that did odd things, so that older versions would keep running. When they were working on Copland, I thought it was nuts to write a new OS from scratch, but still do reimplement the hacks so that everything would still run. Good thing it all went away.

I read a lot of these type of predictions, often you see some people say it's a great time to buy, and others it's a bad time to buy, and generally you will only remember the people who guessed correctly, so none of this helps you today at all. Every time those people who got lucky will appear in some future ad or article making a new prediction, and generally fewer will be lucky twice. In the long run, everyone is likely wrong.

I think it's unwinding the QE it did since 2008.

Enjoy it while you can, Moore v. Harper is up for debate in the Supreme Court this year, and it appears to be in favor of this (insane) idea that voting is no longer a right.

Me as well, I refuse to give Adobe any money if I can use Affinity's products.

Problem is the computing power necessary to implement those bigger models increases by some large value as well. An 81MP images is 81 times bigger than a 1MP image. Not sure exactly how the generation is implemented, but I am sure it probably becomes unaffordable if its linear.

My normal artwork (digital) is 9000x9000 pixels up to about 14400x14400. Good luck getting any of them to make something bigger than 1024x1024. Sure, you can use Gigapixel to upres things, but in the end it's basically impossible to make a decent artwork using ML at the moment. I only use Dall-E for offbeat ideas for geometric abstractions, so far Stable Diffusion sucks for that purpose.

Of course when you upload at no one wants you to do the full resolution image; at 1024x1024 you might be able to fool people with something ML made.

I believe most of these models only use 1k images so they really can't make anything bigger yet. Maybe someday but not today.

The prompt example shows "photrorealistic" which is unlikely to work well.

I wanted to get a PhD in CS at 40 but was convinced by a local professor of CS that it was a waste of everyone's time and there were no jobs to be had in the academic world anyway. Shortly thereafter Google started hiring them by the truckload. I wish I had ignored him.

I did have a 4 decade career as a programmer, but who knows what might have happened if I had taken the time mid-career to do the doctorate.

I have been able to do some algorithmic work in my 4 decades (mostly lately for generative art) but most of my career was How So I Ship This Complicated App With Too Few People In So Little Time. This usually requires creative thinking and experience, but rarely requires advanced knowledge of many of the topics mentioned. I never had a CS education so I had to learn things myself, but it still makes sense to me today to understand complex topics; even if you never use them, being able to understand such things is a benefit to the complex things you will have to do.

I was just trying to figure exactly this, but gave up as I kept hitting a wall. Glad someone smarter than me worked this out.

That is the point of the comment. No matter how bad it gets, biological creatures will adapt as they always have while lots of existing species will become extinct; none of this is new. The question is can people adapt and remain what we think of as people in what we think of as our civilization. Given the current horrific weather (droughts/floods/heat/cold/storms/fires etc) how will humanity adapt without seriously changing what we think of as life. Perhaps we can live underground, or all move to the poles, or try a new planet somewhere. But if weather continues to get worse and worse eventually fewer and fewer people will survive, and we might become like the majority of species that no longer exist.

In even the worst case the Earth will be fine. People, not so much.

In many of my employers over the years where security mattered, it was rarely important enough to spend much money on. One place (covered by HIPAA's rather toothless laws) I pointed out how insecure all of our servers & databases were (single password, known by all), and the only response I got was "we pass our audits, and anyway we trust all of our employees". Groan.

An earlier employer had a single person in charge of security for a company with 50000+ customer investment accounts. Oh and the one that was there when I started was eventually discovered to have two full time jobs, which worked because he had unlimited vacations. After that person was fired, the replacement did nothing but run a few scripts every day, and our databases did not encrypt anything, and they argued for months on whether to buy disk encryption software instead of the just encrypting credit card numbers (which meant various applications had to be modified and no one wanted to pay for that work).

So if Patreon dumped their entire security team (or just one part, it's not clear) makes me reminisce about stupidities... not much has changed.

They should name one Linus.

I retired last year after 39.5 years as a working programmer. I started with Fortran, used C, C++, Java, Objective-C (both modern and NeXT variety), Javascript, lots of assembly languages and the last 6 years, Swift; also dabbled in other languages.

Was always a full time programmer no matter my title and other duties. The key is always learning new things, which is a minefield since so much stuff winds up losing popularity. Most of what I worked on was targeted at people, even server side stuff always had a client as well. I also learned a lot about dealing with people, deciphering and untangling requirements, and how whatever industry I worked in functioned. I only retired because I was tired of working 60 hrs/week and endless meetings.

Compared to today there were only a handful of people programming when I started, so people with 4 decades of experience are still rare.

Sad someone out there will assume this is for real and plan to change their interviewing to this. I've been in interviews where you had to talk with every single person on the team (8, half of which didn't show); a gang interview with me in the middle for several hours; 4 separate interviews with multiple people who all asked complicated puzzle questions; and a place where every team member talked about how they did a part of every job (architect designed classes, lead designed methods, and programmers filled them out), etc. Insanity in interviews knows no bounds.

Two of the best programmers I ever worked with I interviewed over the phone for an hour each and that was it.

Is it really random or just a random set of prerecorded locations? After about 20 or so images I stopped seeing pictures so I refreshed the page - and got a location I had seen previously, which should be impossible.

A rather unique take on spreadsheets that would wind up influencing others to try experimentations on the spreadsheet idea, but all were ultimately killed by Excel, as what happened to us. Search "trapeze spreadsheet" on google if you care to know more.

They wanted to start a manufacturing equipment business like eBay, but for factories, which itself was rather silly given how difficult shipping machines would be. They wanted a preliminary design for a 3 month period, but took 3 months to negotiate the contract, so in the end I worked without any input. I think they were completely clueless about the internet or how difficult such a business would be to run, or even market.

Swift being open source makes sense that the book is too. Hopefully more and better examples will be added.