> also thinking of rendering the back button (like we had in old days before the one small bottom bar)
This still exists in Android though it isn't the default. The first thing I do with any new Android phone is switch on "Buttons" instead of "Swipe gestures" in the OS Navigation bar settings.
The linked proof has been in the works for 12 years and was expected to be worked on for at least another year and was announced early in light of the other GRH announcement.
I don't think anything fundamentally has changed except for the tension introduced by the very real possibility that if you have been working as a human on a math proof for a long time that AI pointed toward the problem you have been working on could scoop you at any time before you get your results out.
> I expect the risk vector would primarily be related to OS-level AI tools being given to see whats being rendered on screen and interact with it. Signal can encrypt at rest and between devices, it can't encrypt when actually rendering text to the UI tree.
Yeah, it is likely this.
A decade ago I worked for a company that provided secure communications apps (email, IMs, etc) for both Android and iOS and (as one of the Android developers) I remember pointing out to higher-ups that it was trivial to write an app on Android that would tie into the accessibility APIs Android provided to bypass all of our app's network and storage security (the accessibility APIs would allow you to automatically read the text out of any standard OS text controls of any installed app while running as a background service to provide alternative means of presenting the data, like text to speech, etc).
But instead of providing accessibility, an attacker app could just take all that data and silently upload it somewhere.
This did mean the attacker would have to physically access the phone one time to install the accessibility logger service (or they'd have to phish a user to install it themselves), but after that initial install they could forever bypass the security on our app with absolutely no way for the user to know data was being exfiled unless they are the type that is meticulously auditing which apps are installed on their phone.
Destroying accessibility is unfair to people who need it, but I recommended we add a feature to our app that would default to notifying the user with a clear warning whenever another running app had accessibility read access.
My recommendation never got followed, at least for the remainder of my time at that company (they got bought up by Blackberry about a year later and the office I was working in got shut down).
> SWEs also often forget that most of the software they create was designed to take jobs away from other people in other departments.
Often maybe, but not always.
Avoiding displacing human labor that doesn't want to be displaced is always something I've thought about before any job I've taken.
No putting clankers before humans, no grifts, scams, pushing ads, gambling or incentivizing other forms of addictive behavior.
Fairly simple rules to live by, though it does mean there are entire major companies (and entire market categories of companies) I could never work for in good conscience.
I do find the fact that AI is disrupting us software developers first and foremost seems karmically fitting in general.
> The death of meaning is way more of a tragedy than most people acknowledge.
I agree that death of meaning is a tragedy, but I also think a lot of us can find plenty of meaning beyond our jobs.
In an optimistic theoretical world where AI took my software dev job but I was sustained through some form of UBI I'd spend all my time doing things I do now in my spare time like volunteering at animal rescues and I'd be perfectly happy and find plenty of meaning in that volunteer work.
Alas, I do not think it is likely that that is our future. Politically and economically we've shown no movement toward that.
So if/when AI takes my job I'll probably have to spend less time than I do now volunteering and spend the majority of the rest of my days trying to cobble together a meager living doing whatever I can. Or perhaps I will turn to a life of crime.
Anthropic has a very high valuation, but isn't yet a profitable company.
And it is still entirely possible Anthropic gets wiped out by commoditized open weight models, that may or may not have included Anthropic model distillation as part of their training (which raises the question again of who pays the royalties in that situation as you move further away from the original data but still end up with usable models).
Given the current structure of our economy there is a very real chance AI destroys a lot of value without having any clear long-term value capturing mechanism to offset the value destruction.
To be clear, I'm not defending Anthropic here or saying that it wasn't theft for these companies to ingest all of this data into their models.
> I don't really know either, just my mental health is torn to shreds due to that impending doom on the horizon.
If it makes you feel any better, there are a lot of us in this position. Far more than talk about it.
I mean, I talk about it all the time but the main response I get is people who are somewhat relieved to learn it isn't just them. (Along with a different bunch of people who need to point out I'm being a doomer to suggest that it is remotely possible that things could get pretty ugly for our profession very soon).
It seems like many people are afraid to talk about how shaky the ground is underneath us for fear that mentioning it will manifest the worst outcome. And a lot of other people seem to be in denial that we could even really be at such an inflection point, clinging to the "AI is normal technology" idea.
I fully accept that my view may be overly pessimistic and things won't be as bad as I think, but I really don't understand how the people who are 100% sure everything is going to be just fine are able to maintain that certainty unless to some degree they are self-soothing as much as anything else.
> I didn't know gas town failed. What happened? Did Steve yegge say as much?
He wrote a series of blog posts in which he talked about how Gas Town failed and blamed it on Opus 4.7 changes causing all his workflows to fail. Then he announced Gas City as a rewrite.
At some point after that he seemed to give up on the whole idea and instead just started working on reviving an old graphical MUD style game he made where I guess now he spends $10,000 a month on agents to build a game that currently has 0 concurrent players (4 concurrent 24 hour peak) on Steam:
The game is available from other platforms than Steam so maybe he's got a bit more users than that across the platforms, and his stated plan is to bilk "whales", so maybe that's working out for him? But it still seems like he's lighting massive amounts of money on fire in what appears to be a prolonged state of ai psychosis.
> I guess Android apps are just easy to reverse engineer or something.
Android apps are indeed easier than average to reverse engineer as they are effectively Dalvik (JVM-like) bytecode wrapped in a zip file format, and Dalvik/JVM style bytecode is among the easiest compiled/shipped executable formats to decompile to a usable source code representation.
More often than not the only protection they have is symbol/function name obfuscation, which makes things a bit tedious for human reverse engineering but which doesn't slow LLMs down much.
This still exists in Android though it isn't the default. The first thing I do with any new Android phone is switch on "Buttons" instead of "Swipe gestures" in the OS Navigation bar settings.
reply