Em Adespoton

If recommendations are being provided to me as a service and the algorithm that goes into it is relatively transparent, I have no issues.

If advertising is based on the value an advertiser sees in the product being advertised, I have no problem.

If I’m the product being sold or an ad distribution network is involved, I’ve got a problem.

Perfect Dad joke material….

TIL acetaminophen isn’t a worldwide thing :)

The whole Tylenol thing is because they’re the ones who originally patented it. Same for Advil and ibuprofen.

Wouldn’t their patch embeddings return different results depending on the visual boundaries? They don’t appear to use overlap redundancy; this means it’s going to be significantly less resource intensive, but the chance of losing significant signals in the image to text translation surely must be inversely high?

Makes sense, as actual AI research is based in applied mathematics and data/signal modelling. And the Chinese education system has trained students in those areas ruthlessly over the past 40 years.

So combine large population base with education system focused on the core competencies required for AI studies, and you’re going to get a majority of the talent coming from that system.

All this really does is show how flawed the current concept of copyright is. But at some point, a huge corpus of images owned by other people was assembled to create a derivative work (the training corpus).

Nope. Bill left MS in 2008 and Windows 7 came out in 2009.

Also the joke left out Windows 10x, AKA 11.

And for some reason, it includes NT and Win2k, but leaves out all the other Server versions (2003 through 23H2).

I spent multiple years learning a skillset which put me into an employment position. Of those jobs I had as an employee 20 years ago, almost all of them were mostly done by machine learning systems a decade later. But that was OK, because I kept on learning and moving ahead of the trend, leaving the learned,boring stuff to automation while I learned new things to give my company a competitive advantage.

I don’t think I could ever work a career where the job I was hired for was my employment until I left.

Why only 4K? We have 8K monitors now.

In which case… they’re both screwed.

They’ll likely have a better chance of pulling it off than America’s push to “delete China”.

One day soon, AI will be cheap enough and good enough that people will produce YouTube videos and AI will summarize the videos into well written, succinct articles with appropriate images and cited sources.

Yes they do… including not holding a charge when the differential drops too far.

The real wins are in battery-backed capacitors. Charge the caps fast, then let them keep the batteries topped up.

One of the first things MS did after buying Mojang was to slap Azure AD on it for account management; and it’s been a number of years now since they switched to that being the only way to authenticate to Minecraft.

This has definitely been the frog boiling strategy at Microsoft for a decade or so. It’s likely a big part of why Windows 11 exits, too.

Was just going to say… my phone has 512GB storage and can do direct WiFi file transfer to my computer without a hotspot. All without using the mobile hotspot feature.

Copyright law doesn’t actually care about commercial use. It probably should, but it doesn’t.

Is the TL;DRbot fair use?

This is a really interesting point; I tried flipping it on its head and the reasoning became even more obvious:

My thought was: “surely we can take advantage of relativistic effects to keep time at a slower pace locally but have it take a short enough time in the referent timeframe.” But in this case, there is a very obvious floor we’re working with: absolute zero. Because making things go relatively faster means making the other things go comparatively slower, and 0 is as slow as you can go. If subatomic particles have no movement, there’s nothing to measure, literally.

As a result, there is a very specific bound on timekeeping measurements no matter how you try to finesse things, with the amount of energy required to make minor improvements ramping up exponentially as that floor is approached.

In order to get around this, we’d have to come up with a different way to do error correction and results measurement, and I’m not sure there is one.

I can see it from both sides. My gmail accounts (regular and throwaway) were roughly my fourth generation email addresses. I got my first email address in 1990. It was tied directly to an educational institution. When I switched institutions, I switched email addresses, and around that time got an ISP email address as well. Non-educational emails went to my ISP address and anything educational related went to my new edu address; everyone in edu circles knew to switch addresses because my .plan file associated with my old account advised them it was closed and what my new one was.

Eventually, I realized that neither my ISP nor edu institution would be with me forever, so I switched everything over to an email redirect service with Yahoo and Hotmail throwaway addresses for stuff that needed an account that was neither professional nor personal.

Then along came Google, Yahoo imploded, Hotmail got bought by Microsoft, and my email redirect service went out of business as the dot com bubble burst.

Oh, and I changed jobs which required moving which meant switching ISPs.

So GMail was a lifeline because I set all my other accounts to both forward to gmail AND set autoresponders informing the sender of my new address.

Of course, that happened 19 years ago. Back then, there were no SMS authentications, no real life accounts tied irrevocably to an email address. My eBay and PayPal accounts just needed an address update, and pretty much everyone else hadn’t got to the point where email address was even an option on a registration form.

That said, I recently did some email address shuffling, and all the accounts that really matter got switched relatively painlessly; I have a password manager, and part of changing addresses involves going through every entry in my password manager (which is already helpfully divided into personal, professional and throwaway) to update addresses as appropriate.

Everyone else gets the same autorespond and redirect treatment for a year. After that, anyone I’ve missed will have to locate me via someone else.

Of course, I’ve also maintained a PGP key since 1993 that has my chain of email addresses associated with it, so anyone who knows my key can just look up my current email address. It’s really the only thing I use that key for anymore. But there’s a very limited set of people that would even think to look me up by PGP, or even save a copy of my public key and remember the key exchange I use.

Google previously proposed putting restrictions on the functionality of this API for security reasons, potentially impacting the effectiveness of ad-blockers across all Chromium-based browsers including Chrome and Microsoft Edge.

Think antitrust may have had something to do with their change of heart?

The laughability of preventing content filtering for security reasons should have been so obvious that not even Google could argue that one with a straight face.